qemu_capabilities.c 166.4 KB
Newer Older
1 2 3
/*
 * qemu_capabilities.c: QEMU capabilities generation
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24
 */

#include <config.h>

#include "qemu_capabilities.h"
25
#include "viralloc.h"
26
#include "vircrypto.h"
27
#include "virlog.h"
28
#include "virerror.h"
E
Eric Blake 已提交
29
#include "virfile.h"
30
#include "virfilecache.h"
31 32
#include "virpidfile.h"
#include "virprocess.h"
33
#include "cpu/cpu.h"
34
#include "cpu/cpu_x86.h"
35
#include "domain_conf.h"
36
#include "vircommand.h"
37
#include "virbitmap.h"
38
#include "virnodesuspend.h"
39
#include "virnuma.h"
40
#include "virhostcpu.h"
41
#include "qemu_monitor.h"
42
#include "virstring.h"
43
#include "qemu_hostdev.h"
44
#include "qemu_domain.h"
45
#define LIBVIRT_QEMU_CAPSPRIV_H_ALLOW
46
#include "qemu_capspriv.h"
47
#include "qemu_qapi.h"
48

49
#include <fcntl.h>
50 51 52
#include <sys/stat.h>
#include <unistd.h>
#include <sys/wait.h>
53
#include <stdarg.h>
54
#include <sys/utsname.h>
55

56 57 58 59
#if WITH_CAPNG
# include <cap-ng.h>
#endif

60 61
#define VIR_FROM_THIS VIR_FROM_QEMU

62 63
VIR_LOG_INIT("qemu.qemu_capabilities");

64 65 66 67
/* While not public, these strings must not change. They
 * are used in domain status files which are read on
 * daemon restarts
 */
68
VIR_ENUM_IMPL(virQEMUCaps, QEMU_CAPS_LAST,
69 70
              /* 0 */
              "kqemu",
71 72 73 74 75
              "vnc-colon",
              "no-reboot",
              "drive",
              "drive-boot",

76 77
              /* 5 */
              "name",
78 79 80 81 82
              "uuid",
              "domid",
              "vnet-hdr",
              "migrate-kvm-stdio",

83 84
              /* 10 */
              "migrate-qemu-tcp",
85 86 87 88 89
              "migrate-qemu-exec",
              "drive-cache-v2",
              "kvm",
              "drive-format",

90 91
              /* 15 */
              "vga",
92 93 94 95 96
              "0.10",
              "pci-device",
              "mem-path",
              "drive-serial",

97 98
              /* 20 */
              "xen-domid",
99 100 101 102 103
              "migrate-qemu-unix",
              "chardev",
              "enable-kvm",
              "monitor-json",

104 105
              /* 25 */
              "balloon",
106 107 108 109 110
              "device",
              "sdl",
              "smp-topology",
              "netdev",

111 112
              /* 30 */
              "rtc",
113
              "vhost-net",
114 115 116 117
              "rtc-td-hack",
              "no-hpet",
              "no-kvm-pit",

118 119
              /* 35 */
              "tdf",
120 121 122 123 124
              "pci-configfd",
              "nodefconfig",
              "boot-menu",
              "enable-kqemu",

125 126
              /* 40 */
              "fsdev",
127 128 129 130 131
              "nesting",
              "name-process",
              "drive-readonly",
              "smbios-type",

132 133
              /* 45 */
              "vga-qxl",
134 135 136 137 138
              "spice",
              "vga-none",
              "migrate-qemu-fd",
              "boot-index",

139 140
              /* 50 */
              "hda-duplex",
141 142 143 144 145
              "drive-aio",
              "pci-multibus",
              "pci-bootindex",
              "ccid-emulated",

146 147
              /* 55 */
              "ccid-passthru",
148 149 150 151
              "chardev-spicevmc",
              "device-spicevmc",
              "virtio-tx-alg",
              "device-qxl-vga",
152

153 154
              /* 60 */
              "pci-multifunction",
155
              "virtio-blk-pci.ioeventfd",
M
Michal Privoznik 已提交
156
              "sga",
157 158
              "virtio-blk-pci.event_idx",
              "virtio-net-pci.event_idx",
159

160 161
              /* 65 */
              "cache-directsync",
162 163 164 165 166
              "piix3-usb-uhci",
              "piix4-usb-uhci",
              "usb-ehci",
              "ich9-usb-ehci1",

167 168
              /* 70 */
              "vt82c686b-usb-uhci",
169 170
              "pci-ohci",
              "usb-redir",
M
Marc-André Lureau 已提交
171
              "usb-hub",
172
              "no-shutdown",
173

174 175
              /* 75 */
              "cache-unsafe",
176
              "rombar",
J
Jim Fehlig 已提交
177
              "ich9-ahci",
178
              "no-acpi",
179
              "fsdev-readonly",
180

181 182
              /* 80 */
              "virtio-blk-pci.scsi",
183
              "blk-sg-io",
O
Osier Yang 已提交
184
              "drive-copy-on-read",
185
              "cpu-host",
186
              "fsdev-writeout",
187

188 189
              /* 85 */
              "drive-iotune",
190
              "system_wakeup",
191
              "scsi-disk.channel",
192
              "scsi-block",
193
              "transaction",
194

195 196
              /* 90 */
              "block-job-sync",
197
              "block-job-async",
198
              "scsi-cd",
199
              "ide-cd",
200
              "no-user-config",
M
Marc-André Lureau 已提交
201

202 203
              /* 95 */
              "hda-micro",
204
              "dump-guest-memory",
G
Gerd Hoffmann 已提交
205
              "nec-usb-xhci",
206
              "virtio-s390",
207
              "balloon-event",
M
Marc-André Lureau 已提交
208

209 210
              /* 100 */
              "bridge",
211 212
              "lsi",
              "virtio-scsi-pci",
V
Viktor Mihajlovski 已提交
213
              "blockio",
214
              "disable-s3",
R
Richa Marwaha 已提交
215

216 217
              /* 105 */
              "disable-s4",
218
              "usb-redir.filter",
219 220
              "ide-drive.wwn",
              "scsi-disk.wwn",
221
              "seccomp-sandbox",
222

223 224
              /* 110 */
              "reboot-timeout",
225
              "dump-guest-core",
226
              "seamless-migration",
227
              "block-commit",
228
              "vnc",
229

230 231
              /* 115 */
              "drive-mirror",
232 233
              "usb-redir.bootindex",
              "usb-host.bootindex",
234
              "blockdev-snapshot-sync",
235 236
              "qxl",

237 238
              /* 120 */
              "VGA",
239 240 241
              "cirrus-vga",
              "vmware-svga",
              "device-video-primary",
242
              "s390-sclp",
243

244 245
              /* 125 */
              "usb-serial",
G
Guannan Ren 已提交
246
              "usb-net",
247
              "add-fd",
248
              "nbd-server",
249 250
              "virtio-rng",

251 252
              /* 130 */
              "rng-random",
253
              "rng-egd",
O
Olivia Yin 已提交
254 255
              "virtio-ccw",
              "dtb",
256
              "megasas",
J
Ján Tomko 已提交
257

258 259
              /* 135 */
              "ipv6-migration",
260
              "machine-opt",
L
Li Zhang 已提交
261
              "machine-usb-opt",
S
Stefan Berger 已提交
262 263
              "tpm-passthrough",
              "tpm-tis",
264

265 266
              /* 140 */
              "nvram",
H
Han Cheng 已提交
267 268 269 270 271
              "pci-bridge",
              "vfio-pci",
              "vfio-pci.bootindex",
              "scsi-generic",

272 273
              /* 145 */
              "scsi-generic.bootindex",
274
              "mem-merge",
275
              "vnc-websocket",
O
Osier Yang 已提交
276
              "drive-discard",
277
              "mlock",
278

279 280
              /* 150 */
              "vnc-share-policy",
281
              "device-del-event",
282
              "dmi-to-pci-bridge",
283 284
              "i440fx-pci-hole64-size",
              "q35-pci-hole64-size",
285

286 287
              /* 155 */
              "usb-storage",
288
              "usb-storage.removable",
289
              "virtio-mmio",
290
              "ich9-intel-hda",
291
              "kvm-pit-lost-tick-policy",
292

293 294
              /* 160 */
              "boot-strict",
295 296
              "pvpanic",
              "enable-fips",
297 298
              "spice-file-xfer-disable",
              "spiceport",
L
Li Zhang 已提交
299

300 301
              /* 165 */
              "usb-kbd",
302
              "host-pci-multidomain",
303
              "msg-timestamp",
304
              "active-commit",
305
              "change-backing-file",
306

307 308
              /* 170 */
              "memory-backend-ram",
309
              "numa",
310
              "memory-backend-file",
311
              "usb-audio",
312
              "rtc-reset-reinjection",
313

314 315
              /* 175 */
              "splash-timeout",
J
John Ferlan 已提交
316
              "iothread",
317
              "migrate-rdma",
318
              "ivshmem",
319
              "drive-iotune-max",
320

321 322
              /* 180 */
              "VGA.vgamem_mb",
323 324 325
              "vmware-svga.vgamem_mb",
              "qxl.vgamem_mb",
              "qxl-vga.vgamem_mb",
326
              "pc-dimm",
327

328 329
              /* 185 */
              "machine-vmport-opt",
330 331
              "aes-key-wrap",
              "dea-key-wrap",
M
Michal Privoznik 已提交
332
              "pci-serial",
333
              "aarch64-off",
334

335 336
              /* 190 */
              "vhost-user-multiqueue",
337
              "migration-event",
338
              "gpex-pcihost",
339
              "ioh3420",
340
              "x3130-upstream",
341

342 343
              /* 195 */
              "xio3130-downstream",
344
              "rtl8139",
345
              "e1000",
346
              "virtio-net",
347
              "gic-version",
348

349 350
              /* 200 */
              "incoming-defer",
M
Marc-André Lureau 已提交
351
              "virtio-gpu",
352
              "virtio-gpu.virgl",
353 354 355
              "virtio-keyboard",
              "virtio-mouse",

356 357
              /* 205 */
              "virtio-tablet",
358
              "virtio-input-host",
359
              "chardev-file-append",
360 361
              "ich9-disable-s3",
              "ich9-disable-s4",
362

363 364
              /* 210 */
              "vserport-change-event",
365
              "virtio-balloon-pci.deflate-on-oom",
366
              "mptsas1068",
367
              "spice-gl",
368 369
              "qxl.vram64_size_mb",

370 371
              /* 215 */
              "qxl-vga.vram64_size_mb",
372
              "chardev-logfile",
373
              "debug-threads",
374
              "secret",
375
              "pxb",
376

377 378
              /* 220 */
              "pxb-pcie",
379
              "device-tray-moved-event",
380
              "nec-usb-xhci-ports",
381
              "virtio-scsi-pci.iothread",
382
              "name-guest",
383

384 385
              /* 225 */
              "qxl.max_outputs",
386
              "qxl-vga.max_outputs",
387
              "spice-unix",
388
              "drive-detect-zeroes",
B
Boris Fiuczynski 已提交
389
              "tls-creds-x509",
390

391 392
              /* 230 */
              "display",
J
Ján Tomko 已提交
393
              "intel-iommu",
M
Michal Privoznik 已提交
394
              "smm",
395
              "virtio-pci-disable-legacy",
396
              "query-hotpluggable-cpus",
397

398 399
              /* 235 */
              "virtio-net.rx_queue_size",
400
              "machine-iommu",
401
              "virtio-vga",
402
              "drive-iotune-max-length",
403 404
              "ivshmem-plain",

405 406
              /* 240 */
              "ivshmem-doorbell",
407
              "query-qmp-schema",
408
              "gluster.debug_level",
409
              "vhost-scsi",
410
              "drive-iotune-group",
411

412 413
              /* 245 */
              "query-cpu-model-expansion",
414
              "virtio-net.host_mtu",
415
              "spice-rendernode",
416
              "nvdimm",
417
              "pcie-root-port",
418

419 420
              /* 250 */
              "query-cpu-definitions",
421
              "block-write-threshold",
422
              "query-named-block-nodes",
423
              "cpu-cache",
424
              "qemu-xhci",
425

426 427
              /* 255 */
              "kernel-irqchip",
428
              "kernel-irqchip.split",
429
              "intel-iommu.intremap",
430
              "intel-iommu.caching-mode",
431
              "intel-iommu.eim",
432

433 434
              /* 260 */
              "intel-iommu.device-iotlb",
435 436
              "virtio.iommu_platform",
              "virtio.ats",
437
              "loadparm",
438
              "spapr-pci-host-bridge",
439 440

              /* 265 */
441
              "spapr-pci-host-bridge.numa_node",
442 443
              "vnc-multi-servers",
              "virtio-net.tx_queue_size",
444
              "chardev-reconnect",
445
              "virtio-gpu.max_outputs",
J
John Ferlan 已提交
446 447 448

              /* 270 */
              "vxhs",
449
              "virtio-blk.num-queues",
450
              "machine.pseries.resize-hpt",
M
Marc-André Lureau 已提交
451
              "vmcoreinfo",
452
              "spapr-vty",
453 454 455

              /* 275 */
              "sclplmconsole",
456
              "numa.dist",
457
              "disk-share-rw",
458
              "iscsi.password-secret",
459
              "isa-serial",
460 461 462

              /* 280 */
              "pl011",
463
              "machine.pseries.max-cpu-compat",
464
              "dump-completed",
465
              "virtio-gpu-ccw",
466 467 468 469 470
              "virtio-keyboard-ccw",

              /* 285 */
              "virtio-mouse-ccw",
              "virtio-tablet-ccw",
471
              "qcow2-luks",
472
              "pcie-pci-bridge",
473
              "seccomp-blacklist",
474 475 476

              /* 290 */
              "query-cpus-fast",
477
              "disk-write-cache",
478
              "nbd-tls",
479
              "tpm-crb",
480
              "pr-manager-helper",
481 482 483

              /* 295 */
              "qom-list-properties",
484
              "memory-backend-file.discard-data",
485
              "virtual-css-bridge",
486 487
              "virtual-css-bridge.cssid-unrestricted",
              "vfio-ccw",
488 489 490

              /* 300 */
              "sdl-gl",
491
              "screendump_device",
492
              "hda-output",
493
              "blockdev-del",
494
              "vmgenid",
495 496 497

              /* 305 */
              "vhost-vsock",
498
              "chardev-fd-pass",
499
              "tpm-emulator",
500 501
              "mch",
              "mch.extended-tseg-mbytes",
502 503 504

              /* 310 */
              "sev-guest",
505
              "machine.pseries.cap-hpt-max-page-size",
506
              "machine.pseries.cap-htm",
507
              "usb-storage.werror",
508
              "egl-headless",
509 510 511

              /* 315 */
              "vfio-pci.display",
512
              "blockdev",
B
Boris Fiuczynski 已提交
513
              "vfio-ap",
Y
Yi Min Zhao 已提交
514
              "zpci",
515
              "memory-backend-memfd",
516 517 518

              /* 320 */
              "memory-backend-memfd.hugetlb",
519
              "iothread.poll-max-ns",
520
              "machine.pseries.cap-nested-hv",
521 522
              "egl-headless.rendernode",
              "memory-backend-file.align",
523 524 525

              /* 325 */
              "memory-backend-file.pmem",
526
              "nvdimm.unarmed",
527
              "scsi-disk.device_id",
528 529
    );

530

531 532 533 534
struct virQEMUCapsMachineType {
    char *name;
    char *alias;
    unsigned int maxCpus;
535
    bool hotplugCpus;
536
    bool qemuDefault;
537
};
538 539 540 541 542 543 544 545 546 547 548

typedef struct _virQEMUCapsHostCPUData virQEMUCapsHostCPUData;
typedef virQEMUCapsHostCPUData *virQEMUCapsHostCPUDataPtr;
struct _virQEMUCapsHostCPUData {
    /* Only the "info" part is stored in the capabilities cache, the rest is
     * re-computed from other fields and external data sources everytime we
     * probe QEMU or load the cache.
     */
    qemuMonitorCPUModelInfoPtr info;
    /* Host CPU definition reported in domain capabilities. */
    virCPUDefPtr reported;
549 550
    /* Migratable host CPU definition used for updating guest CPU. */
    virCPUDefPtr migratable;
551 552 553 554
    /* CPU definition with features detected by libvirt using virCPUGetHost
     * combined with features reported by QEMU. This is used for backward
     * compatible comparison between a guest CPU and a host CPU. */
    virCPUDefPtr full;
555 556
};

557 558 559 560 561 562
/*
 * Update the XML parser/formatter when adding more
 * information to this struct so that it gets cached
 * correctly. It does not have to be ABI-stable, as
 * the cache will be discarded & repopulated if the
 * timestamp on the libvirtd binary changes.
563 564
 *
 * And don't forget to update virQEMUCapsNewCopy.
565
 */
566
struct _virQEMUCaps {
567
    virObject parent;
568

569
    bool usedQMP;
570
    bool kvmSupportsNesting;
571

572
    char *binary;
573
    time_t ctime;
574
    time_t libvirtCtime;
575

576
    virBitmapPtr flags;
577 578 579

    unsigned int version;
    unsigned int kvmVersion;
580
    unsigned int libvirtVersion;
581
    unsigned int microcodeVersion;
582
    char *package;
583
    char *kernelVersion;
584

585
    virArch arch;
586

587 588
    virDomainCapsCPUModelsPtr kvmCPUModels;
    virDomainCapsCPUModelsPtr tcgCPUModels;
589 590

    size_t nmachineTypes;
591
    struct virQEMUCapsMachineType *machineTypes;
A
Andrea Bolognani 已提交
592 593 594

    size_t ngicCapabilities;
    virGICCapability *gicCapabilities;
595

596 597
    virSEVCapability *sevCapabilities;

598 599
    virQEMUCapsHostCPUData kvmCPU;
    virQEMUCapsHostCPUData tcgCPU;
600 601
};

602 603
struct virQEMUCapsSearchData {
    virArch arch;
604
    const char *binaryFilter;
605 606
};

607

608 609
static virClassPtr virQEMUCapsClass;
static void virQEMUCapsDispose(void *obj);
610

611
static int virQEMUCapsOnceInit(void)
612
{
613
    if (!VIR_CLASS_NEW(virQEMUCaps, virClassForObject()))
614 615 616 617 618
        return -1;

    return 0;
}

619
VIR_ONCE_GLOBAL_INIT(virQEMUCaps);
620

621
static virArch virQEMUCapsArchFromString(const char *arch)
622 623 624 625 626
{
    if (STREQ(arch, "i386"))
        return VIR_ARCH_I686;
    if (STREQ(arch, "arm"))
        return VIR_ARCH_ARMV7L;
627 628
    if (STREQ(arch, "or32"))
        return VIR_ARCH_OR32;
629 630 631 632 633

    return virArchFromString(arch);
}


634
static const char *virQEMUCapsArchToString(virArch arch)
635 636 637
{
    if (arch == VIR_ARCH_I686)
        return "i386";
S
Stefan Schallenberg 已提交
638
    else if (arch == VIR_ARCH_ARMV6L || arch == VIR_ARCH_ARMV7L)
639
        return "arm";
640 641
    else if (arch == VIR_ARCH_OR32)
        return "or32";
642 643 644 645

    return virArchToString(arch);
}

646 647 648

/* Checks whether a domain with @guest arch can run natively on @host.
 */
649
bool
650 651 652
virQEMUCapsGuestIsNative(virArch host,
                         virArch guest)
{
653
    /* host & guest arches match */
654 655 656
    if (host == guest)
        return true;

657
    /* hostarch is x86_64 and guest arch is i686 (needs -cpu qemu32) */
658 659 660
    if (host == VIR_ARCH_X86_64 && guest == VIR_ARCH_I686)
        return true;

661
    /* hostarch is aarch64 and guest arch is armv7l (needs -cpu aarch64=off) */
662 663 664
    if (host == VIR_ARCH_AARCH64 && guest == VIR_ARCH_ARMV7L)
        return true;

665
    /* hostarch and guestarch are both ppc64 */
666 667 668 669 670 671 672
    if (ARCH_IS_PPC64(host) && ARCH_IS_PPC64(guest))
        return true;

    return false;
}


673 674 675 676 677 678 679 680 681
/* Given a host and guest architectures, find a suitable QEMU target.
 *
 * This is meant to be used as a second attempt if qemu-system-$guestarch
 * can't be found, eg. on a x86_64 host you want to use qemu-system-i386,
 * if available, instead of qemu-system-x86_64 to run i686 guests */
static virArch
virQEMUCapsFindTarget(virArch hostarch,
                      virArch guestarch)
{
682 683 684
    if (virQEMUCapsGuestIsNative(hostarch, guestarch))
        guestarch = hostarch;

685 686 687 688 689 690
    /* Both ppc64 and ppc64le guests can use the ppc64 target */
    if (ARCH_IS_PPC64(guestarch))
        guestarch = VIR_ARCH_PPC64;

    return guestarch;
}
691

692

693
static void
694 695
virQEMUCapsSetDefaultMachine(virQEMUCapsPtr qemuCaps,
                             size_t defIdx)
696
{
697
    struct virQEMUCapsMachineType tmp = qemuCaps->machineTypes[defIdx];
698 699 700 701

    memmove(qemuCaps->machineTypes + 1,
            qemuCaps->machineTypes,
            sizeof(qemuCaps->machineTypes[0]) * defIdx);
702 703

    qemuCaps->machineTypes[0] = tmp;
704 705
}

706

707
static char *
708 709
virQEMUCapsFindBinary(const char *format,
                      const char *archstr)
710
{
711 712
    char *ret = NULL;
    char *binary = NULL;
713

714
    if (virAsprintf(&binary, format, archstr) < 0)
715
        return NULL;
716 717 718

    ret = virFindFileInPath(binary);
    VIR_FREE(binary);
719 720 721 722 723 724 725 726 727 728 729
    return ret;
}

static char *
virQEMUCapsFindBinaryForArch(virArch hostarch,
                             virArch guestarch)
{
    char *ret = NULL;
    const char *archstr;
    virArch target;

730 731 732 733 734 735 736 737 738
    /* armv7l guests can only take advantage of KVM on aarch64 hosts by
     * using the qemu-system-aarch64 binary, so look for that one first
     * to avoid using qemu-system-arm (and thus TCG) instead */
    if (hostarch == VIR_ARCH_AARCH64 && guestarch == VIR_ARCH_ARMV7L) {
        archstr = virQEMUCapsArchToString(hostarch);
        if ((ret = virQEMUCapsFindBinary("qemu-system-%s", archstr)) != NULL)
            goto out;
    }

739 740 741 742 743 744 745 746 747 748 749
    /* First attempt: try the guest architecture as it is */
    archstr = virQEMUCapsArchToString(guestarch);
    if ((ret = virQEMUCapsFindBinary("qemu-system-%s", archstr)) != NULL)
        goto out;

    /* Second attempt: try looking up by target instead */
    target = virQEMUCapsFindTarget(hostarch, guestarch);
    if (target != guestarch) {
        archstr = virQEMUCapsArchToString(target);
        if ((ret = virQEMUCapsFindBinary("qemu-system-%s", archstr)) != NULL)
            goto out;
750
    }
751

752
 out:
753 754 755
    return ret;
}

756
static int
757
virQEMUCapsInitGuest(virCapsPtr caps,
758
                     virFileCachePtr cache,
759 760
                     virArch hostarch,
                     virArch guestarch)
761 762
{
    char *binary = NULL;
763
    virQEMUCapsPtr qemuCaps = NULL;
764 765
    int ret = -1;

J
Ján Tomko 已提交
766
    /* Check for existence of base emulator, or alternate base
767 768
     * which can be used with magic cpu choice
     */
769
    binary = virQEMUCapsFindBinaryForArch(hostarch, guestarch);
770

771 772 773 774 775 776 777
    /* RHEL doesn't follow the usual naming for QEMU binaries and ships
     * a single binary named qemu-kvm outside of $PATH instead */
    if (virQEMUCapsGuestIsNative(hostarch, guestarch) && !binary) {
        if (VIR_STRDUP(binary, "/usr/libexec/qemu-kvm") < 0)
            return -1;
    }

778
    /* Ignore binary if extracting version info fails */
779
    if (binary) {
780
        if (!(qemuCaps = virQEMUCapsCacheLookup(cache, binary))) {
781 782 783 784
            virResetLastError();
            VIR_FREE(binary);
        }
    }
785

786
    ret = virQEMUCapsInitGuestFromBinary(caps,
787
                                         binary, qemuCaps,
788 789 790
                                         guestarch);

    VIR_FREE(binary);
791
    virObjectUnref(qemuCaps);
792 793 794 795 796 797 798

    return ret;
}

int
virQEMUCapsInitGuestFromBinary(virCapsPtr caps,
                               const char *binary,
799
                               virQEMUCapsPtr qemuCaps,
800 801 802 803 804 805 806 807
                               virArch guestarch)
{
    virCapsGuestPtr guest;
    virCapsGuestMachinePtr *machines = NULL;
    size_t nmachines = 0;
    int ret = -1;
    bool hasdisksnapshot = false;

808 809 810
    if (!binary)
        return 0;

811
    if (virQEMUCapsGetMachineTypesCaps(qemuCaps, &nmachines, &machines) < 0)
812
        goto cleanup;
813 814 815 816

    /* We register kvm as the base emulator too, since we can
     * just give -no-kvm to disable acceleration if required */
    if ((guest = virCapabilitiesAddGuest(caps,
817
                                         VIR_DOMAIN_OSTYPE_HVM,
818
                                         guestarch,
819 820 821 822
                                         binary,
                                         NULL,
                                         nmachines,
                                         machines)) == NULL)
823
        goto cleanup;
824 825 826 827

    machines = NULL;
    nmachines = 0;

A
Andrea Bolognani 已提交
828 829 830
    /* CPU selection is always available, because all QEMU versions
     * we support can use at least '-cpu host' */
    if (!virCapabilitiesAddGuestFeature(guest, "cpuselection", true, false))
831
        goto cleanup;
832

833
    if (!virCapabilitiesAddGuestFeature(guest, "deviceboot", true, false))
834
        goto cleanup;
835

836
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DISK_SNAPSHOT))
837 838
        hasdisksnapshot = true;

839 840
    if (!virCapabilitiesAddGuestFeature(guest, "disksnapshot", hasdisksnapshot,
                                        false))
841
        goto cleanup;
842

D
Daniel P. Berrange 已提交
843
    if (virCapabilitiesAddGuestDomain(guest,
844
                                      VIR_DOMAIN_VIRT_QEMU,
D
Daniel P. Berrange 已提交
845 846 847 848
                                      NULL,
                                      NULL,
                                      0,
                                      NULL) == NULL)
849
        goto cleanup;
850

851
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
852 853 854 855 856 857
        if (virCapabilitiesAddGuestDomain(guest,
                                          VIR_DOMAIN_VIRT_KVM,
                                          NULL,
                                          NULL,
                                          0,
                                          NULL) == NULL) {
858
            goto cleanup;
D
Daniel P. Berrange 已提交
859
        }
860
    }
861

862 863 864
    if ((ARCH_IS_X86(guestarch) || guestarch == VIR_ARCH_AARCH64) &&
        virCapabilitiesAddGuestFeature(guest, "acpi", true, true) == NULL) {
        goto cleanup;
865 866
    }

A
Andrea Bolognani 已提交
867
    if (ARCH_IS_X86(guestarch) &&
868
        virCapabilitiesAddGuestFeature(guest, "apic", true, false) == NULL) {
869
        goto cleanup;
870
    }
871

872
    if ((guestarch == VIR_ARCH_I686) &&
873 874
        (virCapabilitiesAddGuestFeature(guest, "pae", true, false) == NULL ||
         virCapabilitiesAddGuestFeature(guest, "nonpae", true, false) == NULL))
875
        goto cleanup;
876 877 878

    ret = 0;

879
 cleanup:
880 881 882

    virCapabilitiesFreeMachines(machines, nmachines);

883
    return ret;
884 885 886
}


887
virCPUDefPtr
888
virQEMUCapsProbeHostCPUForEmulator(virArch hostArch,
889 890 891
                                   virQEMUCapsPtr qemuCaps,
                                   virDomainVirtType type)
{
892 893
    return virCPUGetHost(hostArch, VIR_CPU_TYPE_GUEST, NULL,
                         virQEMUCapsGetCPUDefinitions(qemuCaps, type));
894 895 896
}


897 898
virCapsPtr
virQEMUCapsInit(virFileCachePtr cache)
899 900
{
    virCapsPtr caps;
901
    size_t i;
T
Tal Kain 已提交
902
    virArch hostarch = virArchFromHost();
903

T
Tal Kain 已提交
904
    if ((caps = virCapabilitiesNew(hostarch,
905
                                   true, true)) == NULL)
906
        goto error;
907

908
    /* Some machines have problematic NUMA topology causing
909 910 911
     * unexpected failures. We don't want to break the QEMU
     * driver in this scenario, so log errors & carry on
     */
M
Martin Kletzander 已提交
912
    if (virCapabilitiesInitNUMA(caps) < 0) {
913
        virCapabilitiesFreeNUMAInfo(caps);
914
        VIR_WARN("Failed to query host NUMA topology, disabling NUMA capabilities");
915 916
    }

917 918 919
    if (virCapabilitiesInitCaches(caps) < 0)
        VIR_WARN("Failed to get host CPU cache info");

M
Martin Kletzander 已提交
920
    if (!(caps->host.cpu = virCPUProbeHost(caps->host.arch)))
921
        VIR_WARN("Failed to get host CPU");
922

923
    /* Add the power management features of the host */
924
    if (virNodeSuspendGetTargetMask(&caps->host.powerMgmt) < 0)
925 926
        VIR_WARN("Failed to get host power management capabilities");

927 928 929
    /* Add IOMMU info */
    virCapabilitiesHostInitIOMMU(caps);

M
Michal Privoznik 已提交
930
    /* Add huge pages info */
931
    if (virCapabilitiesInitPages(caps) < 0)
M
Michal Privoznik 已提交
932 933
        VIR_WARN("Failed to get pages info");

934 935 936
    /* Add domain migration transport URIs */
    virCapabilitiesAddHostMigrateTransport(caps, "tcp");
    virCapabilitiesAddHostMigrateTransport(caps, "rdma");
937

938 939 940 941
    /* QEMU can support pretty much every arch that exists,
     * so just probe for them all - we gracefully fail
     * if a qemu-system-$ARCH binary can't be found
     */
942
    for (i = 0; i < VIR_ARCH_LAST; i++)
943
        if (virQEMUCapsInitGuest(caps, cache,
T
Tal Kain 已提交
944
                                 hostarch,
945
                                 i) < 0)
946
            goto error;
947 948 949

    return caps;

950
 error:
951
    virObjectUnref(caps);
952 953 954 955
    return NULL;
}


956
struct virQEMUCapsStringFlags {
957 958 959 960 961
    const char *value;
    int flag;
};


962 963
struct virQEMUCapsStringFlags virQEMUCapsCommands[] = {
    { "transaction", QEMU_CAPS_TRANSACTION },
964
    { "block-stream", QEMU_CAPS_BLOCKJOB_ASYNC },
965 966 967 968 969 970 971
    { "dump-guest-memory", QEMU_CAPS_DUMP_GUEST_MEMORY },
    { "query-spice", QEMU_CAPS_SPICE },
    { "block-commit", QEMU_CAPS_BLOCK_COMMIT },
    { "query-vnc", QEMU_CAPS_VNC },
    { "drive-mirror", QEMU_CAPS_DRIVE_MIRROR },
    { "blockdev-snapshot-sync", QEMU_CAPS_DISK_SNAPSHOT },
    { "nbd-server-start", QEMU_CAPS_NBD_SERVER },
972
    { "change-backing-file", QEMU_CAPS_CHANGE_BACKING_FILE },
973
    { "rtc-reset-reinjection", QEMU_CAPS_RTC_RESET_REINJECTION },
974
    { "migrate-incoming", QEMU_CAPS_INCOMING_DEFER },
975
    { "query-hotpluggable-cpus", QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS },
976
    { "query-qmp-schema", QEMU_CAPS_QUERY_QMP_SCHEMA },
977 978 979
    { "query-cpu-model-expansion", QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION },
    { "query-cpu-definitions", QEMU_CAPS_QUERY_CPU_DEFINITIONS },
    { "query-named-block-nodes", QEMU_CAPS_QUERY_NAMED_BLOCK_NODES },
980
    { "query-cpus-fast", QEMU_CAPS_QUERY_CPUS_FAST },
981
    { "qom-list-properties", QEMU_CAPS_QOM_LIST_PROPERTIES },
982
    { "blockdev-del", QEMU_CAPS_BLOCKDEV_DEL },
983 984
};

985 986 987 988
struct virQEMUCapsStringFlags virQEMUCapsMigration[] = {
    { "rdma-pin-all", QEMU_CAPS_MIGRATE_RDMA },
};

989
/* Use virQEMUCapsQMPSchemaQueries for querying parameters of events */
990
struct virQEMUCapsStringFlags virQEMUCapsEvents[] = {
991
    { "MIGRATION", QEMU_CAPS_MIGRATION_EVENT },
992
    { "VSERPORT_CHANGE", QEMU_CAPS_VSERPORT_CHANGE },
993
    { "BLOCK_WRITE_THRESHOLD", QEMU_CAPS_BLOCK_WRITE_THRESHOLD },
994
    { "DUMP_COMPLETED", QEMU_CAPS_DUMP_COMPLETED },
995 996
};

997
struct virQEMUCapsStringFlags virQEMUCapsObjectTypes[] = {
998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
    { "hda-duplex", QEMU_CAPS_HDA_DUPLEX },
    { "hda-micro", QEMU_CAPS_HDA_MICRO },
    { "ccid-card-emulated", QEMU_CAPS_CCID_EMULATED },
    { "ccid-card-passthru", QEMU_CAPS_CCID_PASSTHRU },
    { "piix3-usb-uhci", QEMU_CAPS_PIIX3_USB_UHCI },
    { "piix4-usb-uhci", QEMU_CAPS_PIIX4_USB_UHCI },
    { "usb-ehci", QEMU_CAPS_USB_EHCI },
    { "ich9-usb-ehci1", QEMU_CAPS_ICH9_USB_EHCI1 },
    { "vt82c686b-usb-uhci", QEMU_CAPS_VT82C686B_USB_UHCI },
    { "pci-ohci", QEMU_CAPS_PCI_OHCI },
    { "nec-usb-xhci", QEMU_CAPS_NEC_USB_XHCI },
    { "usb-redir", QEMU_CAPS_USB_REDIR },
    { "usb-hub", QEMU_CAPS_USB_HUB },
    { "ich9-ahci", QEMU_CAPS_ICH9_AHCI },
    { "virtio-blk-s390", QEMU_CAPS_VIRTIO_S390 },
1013
    { "virtio-blk-ccw", QEMU_CAPS_VIRTIO_CCW },
1014
    { "sclpconsole", QEMU_CAPS_DEVICE_SCLPCONSOLE },
1015
    { "lsi53c895a", QEMU_CAPS_SCSI_LSI },
1016
    { "virtio-scsi-pci", QEMU_CAPS_VIRTIO_SCSI },
1017 1018
    { "virtio-scsi-s390", QEMU_CAPS_VIRTIO_SCSI },
    { "virtio-scsi-ccw", QEMU_CAPS_VIRTIO_SCSI },
1019
    { "virtio-scsi-device", QEMU_CAPS_VIRTIO_SCSI },
1020
    { "megasas", QEMU_CAPS_SCSI_MEGASAS },
1021
    { "qxl", QEMU_CAPS_DEVICE_QXL },
1022 1023
    { "sga", QEMU_CAPS_SGA },
    { "scsi-block", QEMU_CAPS_SCSI_BLOCK },
1024 1025 1026
    { "VGA", QEMU_CAPS_DEVICE_VGA },
    { "cirrus-vga", QEMU_CAPS_DEVICE_CIRRUS_VGA },
    { "vmware-svga", QEMU_CAPS_DEVICE_VMWARE_SVGA },
H
Han Cheng 已提交
1027
    { "usb-serial", QEMU_CAPS_DEVICE_USB_SERIAL },
1028
    { "virtio-rng-pci", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1029 1030
    { "virtio-rng-s390", QEMU_CAPS_DEVICE_VIRTIO_RNG },
    { "virtio-rng-ccw", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1031
    { "virtio-rng-device", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1032
    { "rng-random", QEMU_CAPS_OBJECT_RNG_RANDOM },
1033
    { "rng-egd", QEMU_CAPS_OBJECT_RNG_EGD },
1034
    { "spapr-nvram", QEMU_CAPS_DEVICE_NVRAM },
1035
    { "pci-bridge", QEMU_CAPS_DEVICE_PCI_BRIDGE },
1036
    { "vfio-pci", QEMU_CAPS_DEVICE_VFIO_PCI },
1037
    { "i82801b11-bridge", QEMU_CAPS_DEVICE_DMI_TO_PCI_BRIDGE },
1038
    { "usb-storage", QEMU_CAPS_DEVICE_USB_STORAGE },
1039
    { "virtio-mmio", QEMU_CAPS_DEVICE_VIRTIO_MMIO },
1040
    { "ich9-intel-hda", QEMU_CAPS_DEVICE_ICH9_INTEL_HDA },
H
Hu Tao 已提交
1041
    { "pvpanic", QEMU_CAPS_DEVICE_PANIC },
L
Li Zhang 已提交
1042
    { "usb-kbd", QEMU_CAPS_DEVICE_USB_KBD },
1043
    { "memory-backend-ram", QEMU_CAPS_OBJECT_MEMORY_RAM },
1044
    { "memory-backend-file", QEMU_CAPS_OBJECT_MEMORY_FILE },
1045
    { "usb-audio", QEMU_CAPS_OBJECT_USB_AUDIO },
J
John Ferlan 已提交
1046
    { "iothread", QEMU_CAPS_OBJECT_IOTHREAD},
1047
    { "ivshmem", QEMU_CAPS_DEVICE_IVSHMEM },
1048
    { "pc-dimm", QEMU_CAPS_DEVICE_PC_DIMM },
M
Michal Privoznik 已提交
1049
    { "pci-serial", QEMU_CAPS_DEVICE_PCI_SERIAL },
1050
    { "gpex-pcihost", QEMU_CAPS_OBJECT_GPEX},
1051
    { "ioh3420", QEMU_CAPS_DEVICE_IOH3420 },
1052
    { "x3130-upstream", QEMU_CAPS_DEVICE_X3130_UPSTREAM },
1053
    { "xio3130-downstream", QEMU_CAPS_DEVICE_XIO3130_DOWNSTREAM },
1054
    { "rtl8139", QEMU_CAPS_DEVICE_RTL8139 },
1055
    { "e1000", QEMU_CAPS_DEVICE_E1000 },
1056 1057 1058 1059
    { "virtio-net-pci", QEMU_CAPS_DEVICE_VIRTIO_NET },
    { "virtio-net-ccw", QEMU_CAPS_DEVICE_VIRTIO_NET },
    { "virtio-net-s390", QEMU_CAPS_DEVICE_VIRTIO_NET },
    { "virtio-net-device", QEMU_CAPS_DEVICE_VIRTIO_NET },
M
Marc-André Lureau 已提交
1060 1061
    { "virtio-gpu-pci", QEMU_CAPS_DEVICE_VIRTIO_GPU },
    { "virtio-gpu-device", QEMU_CAPS_DEVICE_VIRTIO_GPU },
1062
    { "virtio-vga", QEMU_CAPS_DEVICE_VIRTIO_VGA },
1063 1064 1065 1066 1067 1068
    { "virtio-keyboard-device", QEMU_CAPS_VIRTIO_KEYBOARD },
    { "virtio-keyboard-pci", QEMU_CAPS_VIRTIO_KEYBOARD },
    { "virtio-mouse-device", QEMU_CAPS_VIRTIO_MOUSE },
    { "virtio-mouse-pci", QEMU_CAPS_VIRTIO_MOUSE },
    { "virtio-tablet-device", QEMU_CAPS_VIRTIO_TABLET },
    { "virtio-tablet-pci", QEMU_CAPS_VIRTIO_TABLET },
1069 1070
    { "virtio-input-host-device", QEMU_CAPS_VIRTIO_INPUT_HOST },
    { "virtio-input-host-pci", QEMU_CAPS_VIRTIO_INPUT_HOST },
1071
    { "mptsas1068", QEMU_CAPS_SCSI_MPTSAS1068 },
1072
    { "secret", QEMU_CAPS_OBJECT_SECRET },
1073
    { "pxb", QEMU_CAPS_DEVICE_PXB },
1074
    { "pxb-pcie", QEMU_CAPS_DEVICE_PXB_PCIE },
1075
    { "tls-creds-x509", QEMU_CAPS_OBJECT_TLS_CREDS_X509 },
J
Ján Tomko 已提交
1076
    { "intel-iommu", QEMU_CAPS_DEVICE_INTEL_IOMMU },
1077 1078
    { "ivshmem-plain", QEMU_CAPS_DEVICE_IVSHMEM_PLAIN },
    { "ivshmem-doorbell", QEMU_CAPS_DEVICE_IVSHMEM_DOORBELL },
1079
    { "vhost-scsi", QEMU_CAPS_DEVICE_VHOST_SCSI },
1080
    { "nvdimm", QEMU_CAPS_DEVICE_NVDIMM },
1081
    { "pcie-root-port", QEMU_CAPS_DEVICE_PCIE_ROOT_PORT },
1082
    { "qemu-xhci", QEMU_CAPS_DEVICE_QEMU_XHCI },
1083
    { "spapr-pci-host-bridge", QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
M
Marc-André Lureau 已提交
1084
    { "vmcoreinfo", QEMU_CAPS_DEVICE_VMCOREINFO },
1085
    { "spapr-vty", QEMU_CAPS_DEVICE_SPAPR_VTY },
1086
    { "sclplmconsole", QEMU_CAPS_DEVICE_SCLPLMCONSOLE },
1087
    { "isa-serial", QEMU_CAPS_DEVICE_ISA_SERIAL },
1088
    { "pl011", QEMU_CAPS_DEVICE_PL011 },
1089
    { "virtio-gpu-ccw", QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1090 1091 1092
    { "virtio-keyboard-ccw", QEMU_CAPS_DEVICE_VIRTIO_KEYBOARD_CCW },
    { "virtio-mouse-ccw", QEMU_CAPS_DEVICE_VIRTIO_MOUSE_CCW },
    { "virtio-tablet-ccw", QEMU_CAPS_DEVICE_VIRTIO_TABLET_CCW },
1093
    { "pcie-pci-bridge", QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE },
1094
    { "pr-manager-helper", QEMU_CAPS_PR_MANAGER_HELPER },
1095
    { "virtual-css-bridge", QEMU_CAPS_CCW },
1096
    { "vfio-ccw", QEMU_CAPS_DEVICE_VFIO_CCW },
1097
    { "hda-output", QEMU_CAPS_HDA_OUTPUT },
1098
    { "vmgenid", QEMU_CAPS_DEVICE_VMGENID },
1099
    { "vhost-vsock-device", QEMU_CAPS_DEVICE_VHOST_VSOCK },
1100
    { "mch", QEMU_CAPS_DEVICE_MCH },
1101
    { "sev-guest", QEMU_CAPS_SEV_GUEST },
B
Boris Fiuczynski 已提交
1102
    { "vfio-ap", QEMU_CAPS_DEVICE_VFIO_AP },
Y
Yi Min Zhao 已提交
1103
    { "zpci", QEMU_CAPS_DEVICE_ZPCI },
1104
    { "memory-backend-memfd", QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1105 1106
};

1107
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBalloon[] = {
1108
    { "deflate-on-oom", QEMU_CAPS_VIRTIO_BALLOON_AUTODEFLATE },
1109 1110 1111
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1112 1113
};

1114
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBlk[] = {
1115 1116 1117 1118
    { "ioeventfd", QEMU_CAPS_VIRTIO_IOEVENTFD },
    { "event_idx", QEMU_CAPS_VIRTIO_BLK_EVENT_IDX },
    { "scsi", QEMU_CAPS_VIRTIO_BLK_SCSI },
    { "logical_block_size", QEMU_CAPS_BLOCKIO },
1119
    { "num-queues", QEMU_CAPS_VIRTIO_BLK_NUM_QUEUES },
1120
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1121 1122 1123
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1124
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1125 1126
};

1127
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioNet[] = {
1128 1129
    { "tx", QEMU_CAPS_VIRTIO_TX_ALG },
    { "event_idx", QEMU_CAPS_VIRTIO_NET_EVENT_IDX },
1130
    { "rx_queue_size", QEMU_CAPS_VIRTIO_NET_RX_QUEUE_SIZE },
1131
    { "tx_queue_size", QEMU_CAPS_VIRTIO_NET_TX_QUEUE_SIZE },
1132
    { "host_mtu", QEMU_CAPS_VIRTIO_NET_HOST_MTU },
1133 1134 1135
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1136 1137
};

1138
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSpaprPCIHostBridge[] = {
1139 1140 1141
    { "numa_node", QEMU_CAPS_SPAPR_PCI_HOST_BRIDGE_NUMA_NODE },
};

1142
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioSCSI[] = {
1143
    { "iothread", QEMU_CAPS_VIRTIO_SCSI_IOTHREAD },
1144 1145 1146
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1147 1148
};

1149
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVfioPCI[] = {
1150
    { "display", QEMU_CAPS_VFIO_PCI_DISPLAY },
1151 1152
};

1153
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSCSIDisk[] = {
1154 1155
    { "channel", QEMU_CAPS_SCSI_DISK_CHANNEL },
    { "wwn", QEMU_CAPS_SCSI_DISK_WWN },
1156
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1157
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1158
    { "device_id", QEMU_CAPS_SCSI_DISK_DEVICE_ID },
1159 1160
};

1161
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIDEDrive[] = {
1162
    { "wwn", QEMU_CAPS_IDE_DRIVE_WWN },
1163
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1164
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1165 1166
};

1167
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsPiix4PM[] = {
1168 1169
    { "disable_s3", QEMU_CAPS_PIIX_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_PIIX_DISABLE_S4 },
1170 1171
};

1172
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBRedir[] = {
1173
    { "filter", QEMU_CAPS_USB_REDIR_FILTER },
1174 1175
};

1176
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsI440FXPCIHost[] = {
1177 1178 1179
    { "pci-hole64-size", QEMU_CAPS_I440FX_PCI_HOLE64_SIZE },
};

1180
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQ35PCIHost[] = {
1181 1182 1183
    { "pci-hole64-size", QEMU_CAPS_Q35_PCI_HOLE64_SIZE },
};

1184
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBStorage[] = {
1185
    { "removable", QEMU_CAPS_USB_STORAGE_REMOVABLE },
1186
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1187
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1188
    { "werror", QEMU_CAPS_USB_STORAGE_WERROR },
1189 1190
};

1191
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsKVMPit[] = {
1192 1193 1194
    { "lost_tick_policy", QEMU_CAPS_KVM_PIT_TICK_POLICY },
};

1195
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVGA[] = {
1196 1197 1198
    { "vgamem_mb", QEMU_CAPS_VGA_VGAMEM },
};

1199
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVmwareSvga[] = {
1200 1201 1202
    { "vgamem_mb", QEMU_CAPS_VMWARE_SVGA_VGAMEM },
};

1203
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQxl[] = {
1204
    { "vgamem_mb", QEMU_CAPS_QXL_VGAMEM },
1205
    { "vram64_size_mb", QEMU_CAPS_QXL_VRAM64 },
1206
    { "max_outputs", QEMU_CAPS_QXL_MAX_OUTPUTS },
1207 1208
};

1209
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioGpu[] = {
1210
    { "virgl", QEMU_CAPS_VIRTIO_GPU_VIRGL },
1211
    { "max_outputs", QEMU_CAPS_VIRTIO_GPU_MAX_OUTPUTS },
1212 1213 1214
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1215 1216
};

1217
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsICH9[] = {
1218 1219 1220 1221
    { "disable_s3", QEMU_CAPS_ICH9_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_ICH9_DISABLE_S4 },
};

1222
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBNECXHCI[] = {
1223 1224 1225
    { "p3", QEMU_CAPS_NEC_USB_XHCI_PORTS },
};

1226
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIntelIOMMU[] = {
1227
    { "intremap", QEMU_CAPS_INTEL_IOMMU_INTREMAP },
1228
    { "caching-mode", QEMU_CAPS_INTEL_IOMMU_CACHING_MODE },
1229
    { "eim", QEMU_CAPS_INTEL_IOMMU_EIM },
1230
    { "device-iotlb", QEMU_CAPS_INTEL_IOMMU_DEVICE_IOTLB },
1231 1232
};

1233 1234 1235 1236
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsVirtualCSSBridge[] = {
    { "cssid-unrestricted", QEMU_CAPS_CCW_CSSID_UNRESTRICTED },
};

1237 1238 1239 1240
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsMCH[] = {
    { "extended-tseg-mbytes", QEMU_CAPS_MCH_EXTENDED_TSEG_MBYTES },
};

1241 1242 1243 1244
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsNVDIMM[] = {
    { "unarmed", QEMU_CAPS_DEVICE_NVDIMM_UNARMED },
};

1245
/* see documentation for virQEMUQAPISchemaPathGet for the query format */
1246
static struct virQEMUCapsStringFlags virQEMUCapsQMPSchemaQueries[] = {
1247
    { "blockdev-add/arg-type/options/+gluster/debug-level", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
1248
    { "blockdev-add/arg-type/+gluster/debug", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
J
John Ferlan 已提交
1249
    { "blockdev-add/arg-type/+vxhs", QEMU_CAPS_VXHS},
1250
    { "blockdev-add/arg-type/+iscsi/password-secret", QEMU_CAPS_ISCSI_PASSWORD_SECRET },
1251
    { "blockdev-add/arg-type/+qcow2/encrypt/+luks/key-secret", QEMU_CAPS_QCOW2_LUKS },
1252
    { "nbd-server-start/arg-type/tls-creds", QEMU_CAPS_NBD_TLS },
1253
    { "screendump/arg-type/device", QEMU_CAPS_SCREENDUMP_DEVICE },
1254
    { "block-commit/arg-type/*top",  QEMU_CAPS_ACTIVE_COMMIT },
1255
    { "query-iothreads/ret-type/poll-max-ns", QEMU_CAPS_IOTHREAD_POLLING },
1256
    { "query-display-options/ret-type/+egl-headless/rendernode", QEMU_CAPS_EGL_HEADLESS_RENDERNODE },
1257 1258
};

1259 1260
typedef struct _virQEMUCapsObjectTypeProps virQEMUCapsObjectTypeProps;
struct _virQEMUCapsObjectTypeProps {
1261
    const char *type;
1262
    struct virQEMUCapsStringFlags *props;
1263
    size_t nprops;
1264
    int capsCondition;
1265 1266
};

1267 1268 1269 1270 1271
typedef int (*virQEMUCapsObjectTypePropsCB)(qemuMonitorPtr mon,
                                            const char *type,
                                            char ***props);

static virQEMUCapsObjectTypeProps virQEMUCapsDeviceProps[] = {
1272 1273
    { "virtio-blk-pci", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1274
      -1 },
1275 1276
    { "virtio-net-pci", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1277
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1278 1279
    { "virtio-scsi-pci", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1280
      QEMU_CAPS_VIRTIO_SCSI },
1281 1282
    { "virtio-blk-ccw", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1283
      QEMU_CAPS_VIRTIO_CCW },
1284 1285
    { "virtio-net-ccw", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1286
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1287 1288
    { "virtio-scsi-ccw", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1289
      QEMU_CAPS_VIRTIO_SCSI },
1290 1291
    { "virtio-blk-s390", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1292
      QEMU_CAPS_VIRTIO_S390 },
1293 1294
    { "virtio-net-s390", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1295
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1296 1297
    { "vfio-pci", virQEMUCapsDevicePropsVfioPCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVfioPCI),
1298
      QEMU_CAPS_DEVICE_VFIO_PCI },
1299
    { "scsi-hd", virQEMUCapsDevicePropsSCSIDisk,
1300
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSCSIDisk),
1301
      -1 },
1302
    { "ide-hd", virQEMUCapsDevicePropsIDEDrive,
1303
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIDEDrive),
1304
      -1 },
1305 1306
    { "PIIX4_PM", virQEMUCapsDevicePropsPiix4PM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsPiix4PM),
1307
      -1 },
1308 1309
    { "usb-redir", virQEMUCapsDevicePropsUSBRedir,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBRedir),
1310
      QEMU_CAPS_USB_REDIR },
1311 1312
    { "i440FX-pcihost", virQEMUCapsDevicePropsI440FXPCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsI440FXPCIHost),
1313
      -1 },
1314 1315
    { "q35-pcihost", virQEMUCapsDevicePropsQ35PCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQ35PCIHost),
1316
      -1 },
1317 1318
    { "usb-storage", virQEMUCapsDevicePropsUSBStorage,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBStorage),
1319
      QEMU_CAPS_DEVICE_USB_STORAGE },
1320 1321
    { "kvm-pit", virQEMUCapsDevicePropsKVMPit,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsKVMPit),
1322
      -1 },
1323 1324
    { "VGA", virQEMUCapsDevicePropsVGA,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVGA),
1325
      QEMU_CAPS_DEVICE_VGA },
1326 1327
    { "vmware-svga", virQEMUCapsDevicePropsVmwareSvga,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVmwareSvga),
1328
      QEMU_CAPS_DEVICE_VMWARE_SVGA },
1329 1330
    { "qxl", virQEMUCapsDevicePropsQxl,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQxl),
1331
      QEMU_CAPS_DEVICE_QXL },
1332 1333
    { "virtio-gpu-pci", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1334
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1335 1336
    { "virtio-gpu-device", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1337
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1338 1339
    { "ICH9-LPC", virQEMUCapsDevicePropsICH9,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsICH9),
1340
      -1 },
1341 1342
    { "virtio-balloon-pci", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1343
      -1 },
1344 1345
    { "virtio-balloon-ccw", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1346
      -1 },
1347 1348
    { "virtio-balloon-device", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1349
      -1 },
1350 1351
    { "nec-usb-xhci", virQEMUCapsDevicePropsUSBNECXHCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBNECXHCI),
1352
      QEMU_CAPS_NEC_USB_XHCI },
1353 1354
    { "intel-iommu", virQEMUCapsDevicePropsIntelIOMMU,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIntelIOMMU),
1355
      QEMU_CAPS_DEVICE_INTEL_IOMMU },
1356 1357
    { "spapr-pci-host-bridge", virQEMUCapsDevicePropsSpaprPCIHostBridge,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSpaprPCIHostBridge),
1358
      QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
1359 1360
    { "virtio-gpu-ccw", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1361
      QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1362 1363 1364
    { "virtual-css-bridge", virQEMUCapsObjectPropsVirtualCSSBridge,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsVirtualCSSBridge),
      QEMU_CAPS_CCW },
1365 1366 1367
    { "mch", virQEMUCapsDevicePropsMCH,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsMCH),
      QEMU_CAPS_DEVICE_MCH },
1368 1369 1370
    { "nvdimm", virQEMUCapsDevicePropsNVDIMM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsNVDIMM),
      QEMU_CAPS_DEVICE_NVDIMM },
1371 1372
};

1373 1374
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendFile[] = {
    { "discard-data", QEMU_CAPS_OBJECT_MEMORY_FILE_DISCARD },
1375
    { "align", QEMU_CAPS_OBJECT_MEMORY_FILE_ALIGN },
1376
    { "pmem", QEMU_CAPS_OBJECT_MEMORY_FILE_PMEM },
1377 1378
};

1379 1380 1381 1382
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendMemfd[] = {
    { "hugetlb", QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB },
};

1383 1384
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsSPAPRMachine[] = {
    { "cap-hpt-max-page-size", QEMU_CAPS_MACHINE_PSERIES_CAP_HPT_MAX_PAGE_SIZE },
1385
    { "cap-htm", QEMU_CAPS_MACHINE_PSERIES_CAP_HTM },
1386
    { "cap-nested-hv", QEMU_CAPS_MACHINE_PSERIES_CAP_NESTED_HV },
1387 1388
};

1389 1390 1391 1392
static virQEMUCapsObjectTypeProps virQEMUCapsObjectProps[] = {
    { "memory-backend-file", virQEMUCapsObjectPropsMemoryBackendFile,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendFile),
      QEMU_CAPS_OBJECT_MEMORY_FILE },
1393 1394 1395
    { "memory-backend-memfd", virQEMUCapsObjectPropsMemoryBackendMemfd,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendMemfd),
      QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1396 1397 1398
    { "spapr-machine", virQEMUCapsObjectPropsSPAPRMachine,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsSPAPRMachine),
      -1 },
1399
};
1400 1401

static void
1402 1403 1404 1405 1406
virQEMUCapsProcessStringFlags(virQEMUCapsPtr qemuCaps,
                              size_t nflags,
                              struct virQEMUCapsStringFlags *flags,
                              size_t nvalues,
                              char *const*values)
1407 1408
{
    size_t i, j;
1409
    for (i = 0; i < nflags; i++) {
1410 1411 1412
        if (virQEMUCapsGet(qemuCaps, flags[i].flag))
            continue;

1413
        for (j = 0; j < nvalues; j++) {
1414
            if (STREQ(values[j], flags[i].value)) {
1415
                virQEMUCapsSet(qemuCaps, flags[i].flag);
1416 1417 1418 1419 1420 1421 1422
                break;
            }
        }
    }
}


1423
int virQEMUCapsGetDefaultVersion(virCapsPtr caps,
1424
                                 virFileCachePtr capsCache,
1425
                                 unsigned int *version)
1426
{
1427
    virQEMUCapsPtr qemucaps;
T
Tal Kain 已提交
1428
    virArch hostarch;
1429
    virCapsDomainDataPtr capsdata;
1430 1431 1432 1433

    if (*version > 0)
        return 0;

T
Tal Kain 已提交
1434
    hostarch = virArchFromHost();
1435 1436 1437
    if (!(capsdata = virCapabilitiesDomainDataLookup(caps,
            VIR_DOMAIN_OSTYPE_HVM, hostarch, VIR_DOMAIN_VIRT_QEMU,
            NULL, NULL))) {
1438
        virReportError(VIR_ERR_INTERNAL_ERROR,
1439
                       _("Cannot find suitable emulator for %s"),
T
Tal Kain 已提交
1440
                       virArchToString(hostarch));
1441 1442 1443
        return -1;
    }

1444
    qemucaps = virQEMUCapsCacheLookup(capsCache, capsdata->emulator);
1445 1446
    VIR_FREE(capsdata);
    if (!qemucaps)
1447 1448
        return -1;

1449
    *version = virQEMUCapsGetVersion(qemucaps);
1450
    virObjectUnref(qemucaps);
1451 1452
    return 0;
}
1453 1454


1455 1456


1457 1458
virQEMUCapsPtr
virQEMUCapsNew(void)
1459
{
1460
    virQEMUCapsPtr qemuCaps;
1461

1462
    if (virQEMUCapsInitialize() < 0)
1463 1464
        return NULL;

1465
    if (!(qemuCaps = virObjectNew(virQEMUCapsClass)))
1466 1467
        return NULL;

1468
    if (!(qemuCaps->flags = virBitmapNew(QEMU_CAPS_LAST)))
1469
        goto error;
1470

1471
    return qemuCaps;
1472

1473
 error:
1474
    virObjectUnref(qemuCaps);
1475
    return NULL;
1476 1477 1478
}


1479
static int
1480 1481
virQEMUCapsHostCPUDataCopy(virQEMUCapsHostCPUDataPtr dst,
                           virQEMUCapsHostCPUDataPtr src)
1482
{
1483 1484
    if (src->info &&
        !(dst->info = qemuMonitorCPUModelInfoCopy(src->info)))
1485 1486
        return -1;

1487 1488
    if (src->reported &&
        !(dst->reported = virCPUDefCopy(src->reported)))
1489 1490
        return -1;

1491 1492 1493 1494
    if (src->migratable &&
        !(dst->migratable = virCPUDefCopy(src->migratable)))
        return -1;

1495 1496 1497 1498
    if (src->full &&
        !(dst->full = virCPUDefCopy(src->full)))
        return -1;

1499 1500 1501 1502
    return 0;
}


1503
static void
1504
virQEMUCapsHostCPUDataClear(virQEMUCapsHostCPUDataPtr cpuData)
1505
{
1506
    qemuMonitorCPUModelInfoFree(cpuData->info);
1507
    virCPUDefFree(cpuData->reported);
1508
    virCPUDefFree(cpuData->migratable);
1509
    virCPUDefFree(cpuData->full);
1510 1511

    memset(cpuData, 0, sizeof(*cpuData));
1512 1513 1514
}


1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
static int
virQEMUCapsSEVInfoCopy(virSEVCapabilityPtr *dst,
                       virSEVCapabilityPtr src)
{
    VIR_AUTOPTR(virSEVCapability) tmp = NULL;

    if (VIR_ALLOC(tmp) < 0 ||
        VIR_STRDUP(tmp->pdh, src->pdh) < 0 ||
        VIR_STRDUP(tmp->cert_chain, src->cert_chain) < 0)
        return -1;

    tmp->cbitpos = src->cbitpos;
    tmp->reduced_phys_bits = src->reduced_phys_bits;

    VIR_STEAL_PTR(*dst, tmp);
    return 0;
}


1534
virQEMUCapsPtr virQEMUCapsNewCopy(virQEMUCapsPtr qemuCaps)
1535
{
1536
    virQEMUCapsPtr ret = virQEMUCapsNew();
1537 1538 1539 1540 1541
    size_t i;

    if (!ret)
        return NULL;

1542
    ret->usedQMP = qemuCaps->usedQMP;
1543
    ret->kvmSupportsNesting = qemuCaps->kvmSupportsNesting;
1544

1545 1546 1547 1548 1549
    if (VIR_STRDUP(ret->binary, qemuCaps->binary) < 0)
        goto error;

    ret->ctime = qemuCaps->ctime;

1550
    virBitmapCopy(ret->flags, qemuCaps->flags);
1551

1552 1553
    ret->version = qemuCaps->version;
    ret->kvmVersion = qemuCaps->kvmVersion;
1554
    ret->microcodeVersion = qemuCaps->microcodeVersion;
1555 1556 1557 1558

    if (VIR_STRDUP(ret->package, qemuCaps->package) < 0)
        goto error;

1559 1560 1561
    if (VIR_STRDUP(ret->kernelVersion, qemuCaps->kernelVersion) < 0)
        goto error;

1562
    ret->arch = qemuCaps->arch;
1563

1564 1565 1566 1567 1568 1569 1570 1571 1572
    if (qemuCaps->kvmCPUModels) {
        ret->kvmCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->kvmCPUModels);
        if (!ret->kvmCPUModels)
            goto error;
    }

    if (qemuCaps->tcgCPUModels) {
        ret->tcgCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->tcgCPUModels);
        if (!ret->tcgCPUModels)
1573
            goto error;
1574 1575
    }

1576 1577
    if (virQEMUCapsHostCPUDataCopy(&ret->kvmCPU, &qemuCaps->kvmCPU) < 0 ||
        virQEMUCapsHostCPUDataCopy(&ret->tcgCPU, &qemuCaps->tcgCPU) < 0)
1578 1579
        goto error;

1580
    if (VIR_ALLOC_N(ret->machineTypes, qemuCaps->nmachineTypes) < 0)
1581
        goto error;
1582
    ret->nmachineTypes = qemuCaps->nmachineTypes;
1583
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1584 1585
        if (VIR_STRDUP(ret->machineTypes[i].name, qemuCaps->machineTypes[i].name) < 0 ||
            VIR_STRDUP(ret->machineTypes[i].alias, qemuCaps->machineTypes[i].alias) < 0)
1586
            goto error;
1587
        ret->machineTypes[i].maxCpus = qemuCaps->machineTypes[i].maxCpus;
1588
        ret->machineTypes[i].hotplugCpus = qemuCaps->machineTypes[i].hotplugCpus;
1589 1590
        ret->machineTypes[i].qemuDefault = qemuCaps->machineTypes[i].qemuDefault;

1591 1592
    }

1593 1594 1595 1596 1597 1598
    if (VIR_ALLOC_N(ret->gicCapabilities, qemuCaps->ngicCapabilities) < 0)
        goto error;
    ret->ngicCapabilities = qemuCaps->ngicCapabilities;
    for (i = 0; i < qemuCaps->ngicCapabilities; i++)
        ret->gicCapabilities[i] = qemuCaps->gicCapabilities[i];

1599 1600 1601 1602 1603
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST) &&
        virQEMUCapsSEVInfoCopy(&ret->sevCapabilities,
                               qemuCaps->sevCapabilities) < 0)
        goto error;

1604 1605
    return ret;

1606
 error:
1607 1608 1609 1610 1611
    virObjectUnref(ret);
    return NULL;
}


1612
void virQEMUCapsDispose(void *obj)
1613
{
1614
    virQEMUCapsPtr qemuCaps = obj;
1615 1616
    size_t i;

1617
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1618 1619
        VIR_FREE(qemuCaps->machineTypes[i].name);
        VIR_FREE(qemuCaps->machineTypes[i].alias);
1620
    }
1621
    VIR_FREE(qemuCaps->machineTypes);
1622

1623 1624
    virObjectUnref(qemuCaps->kvmCPUModels);
    virObjectUnref(qemuCaps->tcgCPUModels);
1625

1626
    virBitmapFree(qemuCaps->flags);
1627

1628
    VIR_FREE(qemuCaps->package);
1629
    VIR_FREE(qemuCaps->kernelVersion);
1630
    VIR_FREE(qemuCaps->binary);
A
Andrea Bolognani 已提交
1631 1632

    VIR_FREE(qemuCaps->gicCapabilities);
1633

1634 1635
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);

1636 1637
    virQEMUCapsHostCPUDataClear(&qemuCaps->kvmCPU);
    virQEMUCapsHostCPUDataClear(&qemuCaps->tcgCPU);
1638 1639
}

1640
void
1641
virQEMUCapsSet(virQEMUCapsPtr qemuCaps,
1642
               virQEMUCapsFlags flag)
1643
{
1644
    ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1645 1646 1647 1648
}


void
1649
virQEMUCapsSetList(virQEMUCapsPtr qemuCaps, ...)
1650 1651 1652 1653
{
    va_list list;
    int flag;

1654
    va_start(list, qemuCaps);
1655
    while ((flag = va_arg(list, int)) < QEMU_CAPS_LAST)
1656
        ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1657
    va_end(list);
1658 1659 1660 1661
}


void
1662
virQEMUCapsClear(virQEMUCapsPtr qemuCaps,
1663
                 virQEMUCapsFlags flag)
1664
{
1665
    ignore_value(virBitmapClearBit(qemuCaps->flags, flag));
1666 1667 1668
}


1669
char *virQEMUCapsFlagsString(virQEMUCapsPtr qemuCaps)
1670
{
1671
    return virBitmapToString(qemuCaps->flags, true, false);
1672 1673 1674 1675
}


bool
1676
virQEMUCapsGet(virQEMUCapsPtr qemuCaps,
1677
               virQEMUCapsFlags flag)
1678
{
J
Ján Tomko 已提交
1679
    return qemuCaps && virBitmapIsBitSet(qemuCaps->flags, flag);
1680
}
1681 1682


D
Daniel P. Berrange 已提交
1683
bool virQEMUCapsHasPCIMultiBus(virQEMUCapsPtr qemuCaps,
1684
                               const virDomainDef *def)
D
Daniel P. Berrange 已提交
1685
{
1686 1687
    /* x86_64 and i686 support PCI-multibus on all machine types
     * since forever */
1688
    if (ARCH_IS_X86(def->os.arch))
D
Daniel P. Berrange 已提交
1689 1690 1691
        return true;

    if (def->os.arch == VIR_ARCH_PPC ||
1692
        ARCH_IS_PPC64(def->os.arch)) {
D
Daniel P. Berrange 已提交
1693 1694 1695 1696 1697
        /*
         * Usage of pci.0 naming:
         *
         *    ref405ep: no pci
         *       taihu: no pci
1698
         *      bamboo: 1.1.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1699 1700
         *       mac99: 2.0.0
         *     g3beige: 2.0.0
1701
         *        prep: 1.4.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1702 1703 1704 1705 1706 1707
         *     pseries: 2.0.0
         *   mpc8544ds: forever
         * virtex-m507: no pci
         *     ppce500: 1.6.0
         */

1708 1709 1710 1711 1712 1713 1714 1715
        /* We do not store the qemu version in domain status XML.
         * Hope the user is using a QEMU new enough to use 'pci.0',
         * otherwise the results of this function will be wrong
         * for domains already running at the time of daemon
         * restart */
        if (qemuCaps->version == 0)
            return true;

D
Daniel P. Berrange 已提交
1716 1717 1718 1719 1720 1721 1722
        if (qemuCaps->version >= 2000000)
            return true;

        if (qemuCaps->version >= 1006000 &&
            STREQ(def->os.machine, "ppce500"))
            return true;

1723 1724 1725
        if (STREQ(def->os.machine, "bamboo") ||
            STREQ(def->os.machine, "mpc8544ds") ||
            STREQ(def->os.machine, "prep")) {
D
Daniel P. Berrange 已提交
1726
            return true;
1727
        }
D
Daniel P. Berrange 已提交
1728 1729 1730 1731

        return false;
    }

1732 1733 1734 1735
    /* S390 supports PCI-multibus. */
    if (ARCH_IS_S390(def->os.arch))
        return true;

1736 1737 1738 1739
    /* If the virt machine, both on ARM and RISC-V, supports PCI,
     * then it also supports multibus */
    if (qemuDomainIsARMVirt(def) ||
        qemuDomainIsRISCVVirt(def)) {
1740
        return true;
1741
    }
1742

D
Daniel P. Berrange 已提交
1743 1744 1745 1746
    return false;
}


1747
const char *virQEMUCapsGetBinary(virQEMUCapsPtr qemuCaps)
1748
{
1749
    return qemuCaps->binary;
1750 1751
}

1752 1753 1754 1755 1756 1757 1758 1759 1760

void
virQEMUCapsSetArch(virQEMUCapsPtr qemuCaps,
                   virArch arch)
{
    qemuCaps->arch = arch;
}


1761
virArch virQEMUCapsGetArch(virQEMUCapsPtr qemuCaps)
1762
{
1763
    return qemuCaps->arch;
1764 1765 1766
}


1767
unsigned int virQEMUCapsGetVersion(virQEMUCapsPtr qemuCaps)
1768
{
1769
    return qemuCaps->version;
1770 1771 1772
}


1773
unsigned int virQEMUCapsGetKVMVersion(virQEMUCapsPtr qemuCaps)
1774
{
1775
    return qemuCaps->kvmVersion;
1776 1777 1778
}


1779 1780 1781 1782 1783 1784
const char *virQEMUCapsGetPackage(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->package;
}


1785 1786
int
virQEMUCapsAddCPUDefinitions(virQEMUCapsPtr qemuCaps,
1787
                             virDomainVirtType type,
1788
                             const char **name,
1789 1790
                             size_t count,
                             virDomainCapsCPUUsable usable)
1791
{
1792
    size_t i;
1793
    virDomainCapsCPUModelsPtr cpus = NULL;
1794

1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808
    if (type == VIR_DOMAIN_VIRT_KVM && qemuCaps->kvmCPUModels)
        cpus = qemuCaps->kvmCPUModels;
    else if (type == VIR_DOMAIN_VIRT_QEMU && qemuCaps->tcgCPUModels)
        cpus = qemuCaps->tcgCPUModels;

    if (!cpus) {
        if (!(cpus = virDomainCapsCPUModelsNew(count)))
            return -1;

        if (type == VIR_DOMAIN_VIRT_KVM)
            qemuCaps->kvmCPUModels = cpus;
        else
            qemuCaps->tcgCPUModels = cpus;
    }
1809 1810

    for (i = 0; i < count; i++) {
1811
        if (virDomainCapsCPUModelsAdd(cpus, name[i], -1, usable, NULL) < 0)
1812
            return -1;
1813
    }
1814

1815 1816 1817 1818
    return 0;
}


1819
virDomainCapsCPUModelsPtr
1820
virQEMUCapsGetCPUDefinitions(virQEMUCapsPtr qemuCaps,
1821
                             virDomainVirtType type)
1822
{
1823
    if (type == VIR_DOMAIN_VIRT_KVM)
1824
        return qemuCaps->kvmCPUModels;
1825
    else
1826
        return qemuCaps->tcgCPUModels;
1827 1828 1829
}


1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840
static virQEMUCapsHostCPUDataPtr
virQEMUCapsGetHostCPUData(virQEMUCapsPtr qemuCaps,
                          virDomainVirtType type)
{
    if (type == VIR_DOMAIN_VIRT_KVM)
        return &qemuCaps->kvmCPU;
    else
        return &qemuCaps->tcgCPU;
}


1841
virCPUDefPtr
1842
virQEMUCapsGetHostModel(virQEMUCapsPtr qemuCaps,
1843 1844
                        virDomainVirtType type,
                        virQEMUCapsHostCPUType cpuType)
1845
{
1846 1847
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1848 1849 1850
    switch (cpuType) {
    case VIR_QEMU_CAPS_HOST_CPU_REPORTED:
        return cpuData->reported;
1851 1852 1853

    case VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE:
        return cpuData->migratable;
1854 1855 1856 1857 1858

    case VIR_QEMU_CAPS_HOST_CPU_FULL:
        /* 'full' is non-NULL only if we have data from both QEMU and
         * virCPUGetHost */
        return cpuData->full ? cpuData->full : cpuData->reported;
1859 1860 1861
    }

    return NULL;
1862 1863 1864
}


1865 1866 1867
static void
virQEMUCapsSetHostModel(virQEMUCapsPtr qemuCaps,
                        virDomainVirtType type,
1868
                        virCPUDefPtr reported,
1869 1870
                        virCPUDefPtr migratable,
                        virCPUDefPtr full)
1871
{
1872 1873
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1874 1875
    cpuData->reported = reported;
    cpuData->migratable = migratable;
1876
    cpuData->full = full;
1877 1878 1879
}


1880 1881 1882 1883 1884 1885
bool
virQEMUCapsIsCPUModeSupported(virQEMUCapsPtr qemuCaps,
                              virCapsPtr caps,
                              virDomainVirtType type,
                              virCPUMode mode)
{
1886 1887
    virDomainCapsCPUModelsPtr cpus;

1888 1889 1890 1891 1892 1893
    switch (mode) {
    case VIR_CPU_MODE_HOST_PASSTHROUGH:
        return type == VIR_DOMAIN_VIRT_KVM &&
               virQEMUCapsGuestIsNative(caps->host.arch, qemuCaps->arch);

    case VIR_CPU_MODE_HOST_MODEL:
1894 1895
        return !!virQEMUCapsGetHostModel(qemuCaps, type,
                                         VIR_QEMU_CAPS_HOST_CPU_REPORTED);
1896 1897

    case VIR_CPU_MODE_CUSTOM:
1898 1899 1900 1901 1902
        if (type == VIR_DOMAIN_VIRT_KVM)
            cpus = qemuCaps->kvmCPUModels;
        else
            cpus = qemuCaps->tcgCPUModels;
        return cpus && cpus->nmodels > 0;
1903 1904 1905 1906 1907 1908 1909 1910 1911

    case VIR_CPU_MODE_LAST:
        break;
    }

    return false;
}


1912 1913 1914
int virQEMUCapsGetMachineTypesCaps(virQEMUCapsPtr qemuCaps,
                                   size_t *nmachines,
                                   virCapsGuestMachinePtr **machines)
1915 1916 1917 1918
{
    size_t i;

    *machines = NULL;
1919
    *nmachines = qemuCaps->nmachineTypes;
1920

1921 1922 1923 1924
    if (*nmachines &&
        VIR_ALLOC_N(*machines, qemuCaps->nmachineTypes) < 0)
        goto error;

1925
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1926 1927
        virCapsGuestMachinePtr mach;
        if (VIR_ALLOC(mach) < 0)
1928
            goto error;
1929
        (*machines)[i] = mach;
1930 1931 1932
        if (qemuCaps->machineTypes[i].alias) {
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].alias) < 0 ||
                VIR_STRDUP(mach->canonical, qemuCaps->machineTypes[i].name) < 0)
1933
                goto error;
1934
        } else {
1935
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].name) < 0)
1936
                goto error;
1937
        }
1938
        mach->maxCpus = qemuCaps->machineTypes[i].maxCpus;
1939 1940
    }

1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978
    /* Make sure all canonical machine types also have their own entry so that
     * /capabilities/guest/arch[@name='...']/machine/text() XPath selects all
     * supported machine types.
     */
    i = 0;
    while (i < *nmachines) {
        size_t j;
        bool found = false;
        virCapsGuestMachinePtr machine = (*machines)[i];

        if (!machine->canonical) {
            i++;
            continue;
        }

        for (j = 0; j < *nmachines; j++) {
            if (STREQ(machine->canonical, (*machines)[j]->name)) {
                found = true;
                break;
            }
        }

        if (!found) {
            virCapsGuestMachinePtr mach;
            if (VIR_ALLOC(mach) < 0)
                goto error;
            if (VIR_INSERT_ELEMENT_COPY(*machines, i, *nmachines, mach) < 0) {
                VIR_FREE(mach);
                goto error;
            }
            if (VIR_STRDUP(mach->name, machine->canonical) < 0)
                goto error;
            mach->maxCpus = machine->maxCpus;
            i++;
        }
        i++;
    }

1979 1980
    return 0;

1981
 error:
1982 1983 1984 1985 1986 1987 1988
    virCapabilitiesFreeMachines(*machines, *nmachines);
    *nmachines = 0;
    *machines = NULL;
    return -1;
}


1989 1990 1991 1992 1993 1994 1995 1996
/**
 * virQEMUCapsGetCanonicalMachine:
 * @qemuCaps: qemu capabilities object
 * @name: machine name
 *
 * Resolves aliased machine names to the actual machine name. If qemuCaps isn't
 * present @name is returned.
 */
1997 1998
const char *virQEMUCapsGetCanonicalMachine(virQEMUCapsPtr qemuCaps,
                                           const char *name)
1999 2000 2001
{
    size_t i;

2002 2003
    if (!name || !qemuCaps)
        return name;
2004

2005
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2006
        if (!qemuCaps->machineTypes[i].alias)
2007
            continue;
2008 2009
        if (STREQ(qemuCaps->machineTypes[i].alias, name))
            return qemuCaps->machineTypes[i].name;
2010 2011 2012 2013
    }

    return name;
}
2014

2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
const char *
virQEMUCapsGetDefaultMachine(virQEMUCapsPtr qemuCaps)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
        if (qemuCaps->machineTypes[i].qemuDefault)
            return qemuCaps->machineTypes[i].name;
    }

    return NULL;
}
2027

2028 2029 2030 2031 2032 2033 2034 2035 2036
int virQEMUCapsGetMachineMaxCpus(virQEMUCapsPtr qemuCaps,
                                 const char *name)
{
    size_t i;

    if (!name)
        return 0;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2037
        if (!qemuCaps->machineTypes[i].maxCpus)
2038
            continue;
2039 2040
        if (STREQ(qemuCaps->machineTypes[i].name, name))
            return qemuCaps->machineTypes[i].maxCpus;
2041 2042 2043 2044 2045 2046
    }

    return 0;
}


2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060
bool virQEMUCapsGetMachineHotplugCpus(virQEMUCapsPtr qemuCaps,
                                      const char *name)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
        if (STREQ_NULLABLE(qemuCaps->machineTypes[i].name, name))
            return qemuCaps->machineTypes[i].hotplugCpus;
    }

    return false;
}


2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084
/**
 * virQEMUCapsSetGICCapabilities:
 * @qemuCaps: QEMU capabilities
 * @capabilities: GIC capabilities
 * @ncapabilities: number of GIC capabilities
 *
 * Set the GIC capabilities for @qemuCaps.
 *
 * The ownership of @capabilities is taken away from the caller, ie. this
 * function will not make a copy of @capabilities, so releasing that memory
 * after it's been called is a bug.
 */
void
virQEMUCapsSetGICCapabilities(virQEMUCapsPtr qemuCaps,
                              virGICCapability *capabilities,
                              size_t ncapabilities)
{
    VIR_FREE(qemuCaps->gicCapabilities);

    qemuCaps->gicCapabilities = capabilities;
    qemuCaps->ngicCapabilities = ncapabilities;
}


2085 2086 2087 2088 2089 2090 2091
virSEVCapabilityPtr
virQEMUCapsGetSEVCapabilities(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->sevCapabilities;
}


2092
static int
2093 2094
virQEMUCapsProbeQMPCommands(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2095 2096 2097 2098 2099 2100 2101
{
    char **commands = NULL;
    int ncommands;

    if ((ncommands = qemuMonitorGetCommands(mon, &commands)) < 0)
        return -1;

2102 2103 2104 2105
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsCommands),
                                  virQEMUCapsCommands,
                                  ncommands, commands);
2106
    virStringListFreeCount(commands, ncommands);
2107

2108 2109 2110 2111
    /* Probe for active commit of qemu 2.1. We don't need to query directly
     * if we have QMP schema support */
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA) &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCK_COMMIT) &&
2112 2113 2114
        qemuMonitorSupportsActiveCommit(mon))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_ACTIVE_COMMIT);

2115 2116 2117 2118 2119
    return 0;
}


static int
2120 2121
virQEMUCapsProbeQMPEvents(virQEMUCapsPtr qemuCaps,
                          qemuMonitorPtr mon)
2122 2123 2124 2125
{
    char **events = NULL;
    int nevents;

2126
    /* we can probe events also from the QMP schema so we can skip this here */
2127
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA))
2128
        return 0;
2129 2130 2131

    if ((nevents = qemuMonitorGetEvents(mon, &events)) < 0)
        return -1;
2132

2133 2134 2135 2136
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsEvents),
                                  virQEMUCapsEvents,
                                  nevents, events);
2137
    virStringListFreeCount(events, nevents);
2138 2139 2140 2141

    return 0;
}

2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170
static int
virQEMUCapsProbeQMPGenericProps(virQEMUCapsPtr qemuCaps,
                                qemuMonitorPtr mon,
                                virQEMUCapsObjectTypeProps *props,
                                size_t nprops,
                                virQEMUCapsObjectTypePropsCB propsGetCB)
{
    int nvalues;
    char **values;
    size_t i;

    for (i = 0; i < nprops; i++) {
        const char *type = props[i].type;
        int cap = props[i].capsCondition;

        if (cap >= 0 && !virQEMUCapsGet(qemuCaps, cap))
            continue;

        if ((nvalues = propsGetCB(mon, type, &values)) < 0)
            return -1;
        virQEMUCapsProcessStringFlags(qemuCaps,
                                      props[i].nprops,
                                      props[i].props,
                                      nvalues, values);
        virStringListFreeCount(values, nvalues);
    }

    return 0;
}
2171

2172
static int
2173
virQEMUCapsProbeQMPDevices(virQEMUCapsPtr qemuCaps,
2174
                           qemuMonitorPtr mon)
2175 2176 2177 2178 2179 2180
{
    int nvalues;
    char **values;

    if ((nvalues = qemuMonitorGetObjectTypes(mon, &values)) < 0)
        return -1;
2181 2182 2183 2184
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsObjectTypes),
                                  virQEMUCapsObjectTypes,
                                  nvalues, values);
2185
    virStringListFreeCount(values, nvalues);
2186

2187 2188 2189 2190 2191 2192
    if (virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsDeviceProps,
                                        ARRAY_CARDINALITY(virQEMUCapsDeviceProps),
                                        qemuMonitorGetDeviceProps) < 0)
        return -1;
2193

2194 2195 2196 2197 2198 2199 2200 2201
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QOM_LIST_PROPERTIES) &&
        virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsObjectProps,
                                        ARRAY_CARDINALITY(virQEMUCapsObjectProps),
                                        qemuMonitorGetObjectProps) < 0)
        return -1;

2202 2203 2204 2205
    return 0;
}


2206 2207 2208 2209 2210 2211 2212 2213 2214 2215
/* Historically QEMU x86 targets defaulted to 'pc' machine type but
 * in future x86_64 might switch to 'q35'. Such a change is considered
 * an ABI break from libvirt's POV. Other QEMU targets may not declare
 * a default machine at all, causing libvirt to use the first reported
 * machine in the list.
 *
 * Here we record a preferred default machine for all arches, so
 * that we're not vulnerable to changes in QEMU defaults or machine
 * list ordering.
 */
2216
static const char *preferredMachines[] =
2217
{
2218 2219
    NULL, /* VIR_ARCH_NONE (not a real arch :) */
    "clipper", /* VIR_ARCH_ALPHA */
S
Stefan Schallenberg 已提交
2220
    "integratorcp", /* VIR_ARCH_ARMV6L */
2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247
    "integratorcp", /* VIR_ARCH_ARMV7L */
    "integratorcp", /* VIR_ARCH_ARMV7B */

    "integratorcp", /* VIR_ARCH_AARCH64 */
    "axis-dev88", /* VIR_ARCH_CRIS */
    "pc", /* VIR_ARCH_I686 */
    NULL, /* VIR_ARCH_ITANIUM (doesn't exist in QEMU any more) */
    "lm32-evr", /* VIR_ARCH_LM32 */

    "mcf5208evb", /* VIR_ARCH_M68K */
    "petalogix-s3adsp1800", /* VIR_ARCH_MICROBLAZE */
    "petalogix-s3adsp1800", /* VIR_ARCH_MICROBLAZEEL */
    "malta", /* VIR_ARCH_MIPS */
    "malta", /* VIR_ARCH_MIPSEL */

    "malta", /* VIR_ARCH_MIPS64 */
    "malta", /* VIR_ARCH_MIPS64EL */
    "or1k-sim", /* VIR_ARCH_OR32 */
    NULL, /* VIR_ARCH_PARISC (no QEMU impl) */
    NULL, /* VIR_ARCH_PARISC64 (no QEMU impl) */

    "g3beige", /* VIR_ARCH_PPC */
    "g3beige", /* VIR_ARCH_PPCLE */
    "pseries", /* VIR_ARCH_PPC64 */
    "pseries", /* VIR_ARCH_PPC64LE */
    "bamboo", /* VIR_ARCH_PPCEMB */

L
Lubomir Rintel 已提交
2248 2249
    "spike_v1.10", /* VIR_ARCH_RISCV32 */
    "spike_v1.10", /* VIR_ARCH_RISCV64 */
2250 2251 2252
    NULL, /* VIR_ARCH_S390 (no QEMU impl) */
    "s390-ccw-virtio", /* VIR_ARCH_S390X */
    "shix", /* VIR_ARCH_SH4 */
L
Lubomir Rintel 已提交
2253

2254 2255 2256 2257 2258
    "shix", /* VIR_ARCH_SH4EB */
    "SS-5", /* VIR_ARCH_SPARC */
    "sun4u", /* VIR_ARCH_SPARC64 */
    "puv3", /* VIR_ARCH_UNICORE32 */
    "pc", /* VIR_ARCH_X86_64 */
L
Lubomir Rintel 已提交
2259

2260 2261
    "sim", /* VIR_ARCH_XTENSA */
    "sim", /* VIR_ARCH_XTENSAEB */
2262
};
2263
verify(ARRAY_CARDINALITY(preferredMachines) == VIR_ARCH_LAST);
2264 2265


2266
static int
2267 2268
virQEMUCapsProbeQMPMachineTypes(virQEMUCapsPtr qemuCaps,
                                qemuMonitorPtr mon)
2269 2270 2271 2272 2273
{
    qemuMonitorMachineInfoPtr *machines = NULL;
    int nmachines = 0;
    int ret = -1;
    size_t i;
2274 2275 2276
    ssize_t defIdx = -1;
    ssize_t preferredIdx = -1;
    const char *preferredMachine = preferredMachines[qemuCaps->arch];
2277 2278

    if ((nmachines = qemuMonitorGetMachines(mon, &machines)) < 0)
2279
        return -1;
2280

2281
    if (VIR_ALLOC_N(qemuCaps->machineTypes, nmachines) < 0)
2282 2283
        goto cleanup;

2284
    for (i = 0; i < nmachines; i++) {
2285
        struct virQEMUCapsMachineType *mach;
2286 2287
        if (STREQ(machines[i]->name, "none"))
            continue;
2288 2289 2290 2291 2292

        mach = &(qemuCaps->machineTypes[qemuCaps->nmachineTypes++]);

        if (VIR_STRDUP(mach->alias, machines[i]->alias) < 0 ||
            VIR_STRDUP(mach->name, machines[i]->name) < 0)
2293
            goto cleanup;
2294 2295

        mach->maxCpus = machines[i]->maxCpus;
2296
        mach->hotplugCpus = machines[i]->hotplugCpus;
2297

2298 2299 2300 2301 2302 2303
        if (preferredMachine &&
            (STREQ_NULLABLE(mach->alias, preferredMachine) ||
             STREQ(mach->name, preferredMachine))) {
            preferredIdx = qemuCaps->nmachineTypes - 1;
        }

2304 2305
        if (machines[i]->isDefault) {
            mach->qemuDefault = true;
2306
            defIdx = qemuCaps->nmachineTypes - 1;
2307
        }
2308
    }
2309

2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322
    /*
     * We'll prefer to use our own historical default machine
     * to avoid mgmt apps seeing semantics changes when QEMU
     * alters its defaults.
     *
     * Our preferred machine might have been compiled out of
     * QEMU at build time though, so we still fallback to honouring
     * QEMU's reported default in that case
     */
    if (preferredIdx == -1)
        preferredIdx = defIdx;
    if (preferredIdx != -1)
        virQEMUCapsSetDefaultMachine(qemuCaps, preferredIdx);
2323 2324 2325

    ret = 0;

2326
 cleanup:
2327
    for (i = 0; i < nmachines; i++)
2328 2329 2330 2331 2332 2333
        qemuMonitorMachineInfoFree(machines[i]);
    VIR_FREE(machines);
    return ret;
}


2334 2335
virDomainCapsCPUModelsPtr
virQEMUCapsFetchCPUDefinitions(qemuMonitorPtr mon)
2336
{
2337 2338 2339
    virDomainCapsCPUModelsPtr models = NULL;
    qemuMonitorCPUDefInfoPtr *cpus = NULL;
    int ncpus = 0;
2340
    size_t i;
2341

2342
    if ((ncpus = qemuMonitorGetCPUDefinitions(mon, &cpus)) < 0)
2343
        return NULL;
2344

2345
    if (!(models = virDomainCapsCPUModelsNew(ncpus)))
2346
        goto error;
2347

2348
    for (i = 0; i < ncpus; i++) {
2349 2350 2351 2352 2353 2354 2355
        virDomainCapsCPUUsable usable = VIR_DOMCAPS_CPU_USABLE_UNKNOWN;

        if (cpus[i]->usable == VIR_TRISTATE_BOOL_YES)
            usable = VIR_DOMCAPS_CPU_USABLE_YES;
        else if (cpus[i]->usable == VIR_TRISTATE_BOOL_NO)
            usable = VIR_DOMCAPS_CPU_USABLE_NO;

2356
        if (virDomainCapsCPUModelsAddSteal(models, &cpus[i]->name, usable,
2357
                                           &cpus[i]->blockers) < 0)
2358
            goto error;
2359 2360 2361 2362 2363 2364
    }

 cleanup:
    for (i = 0; i < ncpus; i++)
        qemuMonitorCPUDefInfoFree(cpus[i]);
    VIR_FREE(cpus);
2365 2366 2367 2368 2369 2370
    return models;

 error:
    virObjectUnref(models);
    models = NULL;
    goto cleanup;
2371 2372
}

2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395

int
virQEMUCapsProbeQMPCPUDefinitions(virQEMUCapsPtr qemuCaps,
                                  qemuMonitorPtr mon,
                                  bool tcg)
{
    virDomainCapsCPUModelsPtr models = NULL;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_DEFINITIONS))
        return 0;

    if (!(models = virQEMUCapsFetchCPUDefinitions(mon)))
        return -1;

    if (tcg || !virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM))
        qemuCaps->tcgCPUModels = models;
    else
        qemuCaps->kvmCPUModels = models;

    return 0;
}


2396 2397
static int
virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps,
2398 2399
                           qemuMonitorPtr mon,
                           bool tcg)
2400
{
2401
    qemuMonitorCPUModelInfoPtr modelInfo = NULL;
2402 2403
    qemuMonitorCPUModelInfoPtr nonMigratable = NULL;
    virHashTablePtr hash = NULL;
2404
    const char *model;
2405
    qemuMonitorCPUModelExpansionType type;
2406 2407
    virDomainVirtType virtType;
    virQEMUCapsHostCPUDataPtr cpuData;
2408
    int ret = -1;
2409 2410

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
2411 2412
        return 0;

2413
    if (tcg || !virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
2414
        virtType = VIR_DOMAIN_VIRT_QEMU;
2415 2416
        model = "max";
    } else {
2417
        virtType = VIR_DOMAIN_VIRT_KVM;
2418 2419 2420
        model = "host";
    }

2421 2422
    cpuData = virQEMUCapsGetHostCPUData(qemuCaps, virtType);

2423 2424 2425 2426 2427 2428 2429 2430 2431 2432
    /* Some x86_64 features defined in cpu_map.xml use spelling which differ
     * from the one preferred by QEMU. Static expansion would give us only the
     * preferred spelling, thus we need to do a full expansion on the result of
     * the initial static expansion to get all variants of all features.
     */
    if (ARCH_IS_X86(qemuCaps->arch))
        type = QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC_FULL;
    else
        type = QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC;

2433 2434
    if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &modelInfo) < 0)
        goto cleanup;
2435 2436

    /* Try to check migratability of each feature. */
2437
    if (modelInfo &&
2438 2439
        qemuMonitorGetCPUModelExpansion(mon, type, model, false,
                                        &nonMigratable) < 0)
2440
        goto cleanup;
2441 2442 2443 2444 2445 2446 2447

    if (nonMigratable) {
        qemuMonitorCPUPropertyPtr prop;
        qemuMonitorCPUPropertyPtr nmProp;
        size_t i;

        if (!(hash = virHashCreate(0, NULL)))
2448
            goto cleanup;
2449

2450 2451
        for (i = 0; i < modelInfo->nprops; i++) {
            prop = modelInfo->props + i;
2452
            if (virHashAddEntry(hash, prop->name, prop) < 0)
2453
                goto cleanup;
2454 2455 2456 2457 2458 2459 2460 2461 2462
        }

        for (i = 0; i < nonMigratable->nprops; i++) {
            nmProp = nonMigratable->props + i;
            if (!(prop = virHashLookup(hash, nmProp->name)) ||
                prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN ||
                prop->type != nmProp->type)
                continue;

2463
            if (prop->value.boolean) {
2464
                prop->migratable = VIR_TRISTATE_BOOL_YES;
2465 2466 2467 2468
            } else if (nmProp->value.boolean) {
                prop->value.boolean = true;
                prop->migratable = VIR_TRISTATE_BOOL_NO;
            }
2469 2470
        }

2471
        modelInfo->migratability = true;
2472 2473
    }

2474
    VIR_STEAL_PTR(cpuData->info, modelInfo);
2475 2476 2477 2478 2479
    ret = 0;

 cleanup:
    virHashFree(hash);
    qemuMonitorCPUModelInfoFree(nonMigratable);
2480
    qemuMonitorCPUModelInfoFree(modelInfo);
2481 2482

    return ret;
2483 2484
}

2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536

/**
 * Get NULL terminated list of features supported by QEMU.
 *
 * Returns -1 on error,
 *          0 on success (@features will be NULL if QEMU does not support this),
 *          1 when @features is filled in, but migratability info is not available.
 */
int
virQEMUCapsGetCPUFeatures(virQEMUCapsPtr qemuCaps,
                          virDomainVirtType virtType,
                          bool migratable,
                          char ***features)
{
    virQEMUCapsHostCPUDataPtr data;
    char **list;
    size_t i;
    size_t n;
    int ret = -1;

    *features = NULL;
    data = virQEMUCapsGetHostCPUData(qemuCaps, virtType);

    if (!data->info)
        return 0;

    if (VIR_ALLOC_N(list, data->info->nprops + 1) < 0)
        return -1;

    n = 0;
    for (i = 0; i < data->info->nprops; i++) {
        qemuMonitorCPUPropertyPtr prop = data->info->props + i;

        if (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO)
            continue;

        if (VIR_STRDUP(list[n++], prop->name) < 0)
            goto cleanup;
    }

    VIR_STEAL_PTR(*features, list);
    if (migratable && !data->info->migratability)
        ret = 1;
    else
        ret = 0;

 cleanup:
    virStringListFree(list);
    return ret;
}


2537 2538
struct tpmTypeToCaps {
    int type;
2539
    virQEMUCapsFlags caps;
2540 2541 2542 2543 2544 2545 2546
};

static const struct tpmTypeToCaps virQEMUCapsTPMTypesToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_TYPE_PASSTHROUGH,
        .caps = QEMU_CAPS_DEVICE_TPM_PASSTHROUGH,
    },
2547 2548 2549 2550
    {
        .type = VIR_DOMAIN_TPM_TYPE_EMULATOR,
        .caps = QEMU_CAPS_DEVICE_TPM_EMULATOR,
    },
2551 2552 2553 2554 2555 2556 2557
};

const struct tpmTypeToCaps virQEMUCapsTPMModelsToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_MODEL_TIS,
        .caps = QEMU_CAPS_DEVICE_TPM_TIS,
    },
2558 2559 2560 2561
    {
        .type = VIR_DOMAIN_TPM_MODEL_CRB,
        .caps = QEMU_CAPS_DEVICE_TPM_CRB,
    },
2562 2563 2564 2565 2566 2567
};

static int
virQEMUCapsProbeQMPTPM(virQEMUCapsPtr qemuCaps,
                       qemuMonitorPtr mon)
{
2568 2569
    int nentries;
    size_t i;
2570
    char **entries = NULL;
S
Stefan Berger 已提交
2571

2572 2573 2574 2575 2576 2577 2578
    if ((nentries = qemuMonitorGetTPMModels(mon, &entries)) < 0)
        return -1;

    if (nentries > 0) {
        for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsTPMModelsToCaps); i++) {
            const char *needle = virDomainTPMModelTypeToString(
                virQEMUCapsTPMModelsToCaps[i].type);
2579
            if (virStringListHasString((const char **)entries, needle))
2580 2581 2582 2583
                virQEMUCapsSet(qemuCaps,
                               virQEMUCapsTPMModelsToCaps[i].caps);
        }
    }
2584
    virStringListFree(entries);
2585 2586 2587 2588 2589 2590 2591 2592

    if ((nentries = qemuMonitorGetTPMTypes(mon, &entries)) < 0)
        return -1;

    if (nentries > 0) {
        for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsTPMTypesToCaps); i++) {
            const char *needle = virDomainTPMBackendTypeToString(
                virQEMUCapsTPMTypesToCaps[i].type);
2593
            if (virStringListHasString((const char **)entries, needle))
2594 2595 2596
                virQEMUCapsSet(qemuCaps, virQEMUCapsTPMTypesToCaps[i].caps);
        }
    }
2597
    virStringListFree(entries);
2598 2599 2600 2601

    return 0;
}

2602

2603
static int
2604 2605
virQEMUCapsProbeQMPKVMState(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2606 2607 2608 2609 2610 2611 2612
{
    bool enabled = false;
    bool present = false;

    if (qemuMonitorGetKVMState(mon, &enabled, &present) < 0)
        return -1;

A
Andrea Bolognani 已提交
2613 2614
    if (present && enabled)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_KVM);
2615 2616 2617 2618

    return 0;
}

2619 2620 2621 2622 2623 2624 2625 2626
struct virQEMUCapsCommandLineProps {
    const char *option;
    const char *param;
    int flag;
};

static struct virQEMUCapsCommandLineProps virQEMUCapsCommandLine[] = {
    { "machine", "mem-merge", QEMU_CAPS_MEM_MERGE },
2627
    { "machine", "vmport", QEMU_CAPS_MACHINE_VMPORT_OPT },
O
Osier Yang 已提交
2628
    { "drive", "discard", QEMU_CAPS_DRIVE_DISCARD },
2629
    { "drive", "detect-zeroes", QEMU_CAPS_DRIVE_DETECT_ZEROES },
2630
    { "realtime", "mlock", QEMU_CAPS_REALTIME_MLOCK },
2631
    { "boot-opts", "strict", QEMU_CAPS_BOOT_STRICT },
2632
    { "boot-opts", "reboot-timeout", QEMU_CAPS_REBOOT_TIMEOUT },
2633
    { "boot-opts", "splash-time", QEMU_CAPS_SPLASH_TIMEOUT },
2634
    { "spice", "disable-agent-file-xfer", QEMU_CAPS_SPICE_FILE_XFER_DISABLE },
2635
    { "msg", "timestamp", QEMU_CAPS_MSG_TIMESTAMP },
2636
    { "numa", NULL, QEMU_CAPS_NUMA },
2637
    { "drive", "throttling.bps-total-max", QEMU_CAPS_DRIVE_IOTUNE_MAX},
2638 2639
    { "machine", "aes-key-wrap", QEMU_CAPS_AES_KEY_WRAP },
    { "machine", "dea-key-wrap", QEMU_CAPS_DEA_KEY_WRAP },
2640
    { "chardev", "append", QEMU_CAPS_CHARDEV_FILE_APPEND },
2641
    { "spice", "gl", QEMU_CAPS_SPICE_GL },
2642
    { "chardev", "logfile", QEMU_CAPS_CHARDEV_LOGFILE },
2643
    { "name", "debug-threads", QEMU_CAPS_NAME_DEBUG_THREADS },
2644
    { "name", "guest", QEMU_CAPS_NAME_GUEST },
2645
    { "spice", "unix", QEMU_CAPS_SPICE_UNIX },
2646
    { "drive", "throttling.bps-total-max-length", QEMU_CAPS_DRIVE_IOTUNE_MAX_LENGTH },
2647
    { "drive", "throttling.group", QEMU_CAPS_DRIVE_IOTUNE_GROUP },
2648
    { "spice", "rendernode", QEMU_CAPS_SPICE_RENDERNODE },
2649
    { "machine", "kernel_irqchip", QEMU_CAPS_MACHINE_KERNEL_IRQCHIP },
2650
    { "machine", "loadparm", QEMU_CAPS_LOADPARM },
2651
    { "vnc", "vnc", QEMU_CAPS_VNC_MULTI_SERVERS },
2652
    { "chardev", "reconnect", QEMU_CAPS_CHARDEV_RECONNECT },
2653
    { "sandbox", "enable", QEMU_CAPS_SECCOMP_SANDBOX },
2654
    { "sandbox", "elevateprivileges", QEMU_CAPS_SECCOMP_BLACKLIST },
2655
    { "chardev", "fd", QEMU_CAPS_CHARDEV_FD_PASS },
2656 2657 2658 2659 2660 2661
};

static int
virQEMUCapsProbeQMPCommandLine(virQEMUCapsPtr qemuCaps,
                               qemuMonitorPtr mon)
{
2662
    bool found = false;
2663 2664 2665 2666 2667 2668 2669
    int nvalues;
    char **values;
    size_t i, j;

    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsCommandLine); i++) {
        if ((nvalues = qemuMonitorGetCommandLineOptionParameters(mon,
                                                                 virQEMUCapsCommandLine[i].option,
2670 2671
                                                                 &values,
                                                                 &found)) < 0)
2672
            return -1;
2673 2674 2675 2676

        if (found && !virQEMUCapsCommandLine[i].param)
            virQEMUCapsSet(qemuCaps, virQEMUCapsCommandLine[i].flag);

2677
        for (j = 0; j < nvalues; j++) {
2678
            if (STREQ_NULLABLE(virQEMUCapsCommandLine[i].param, values[j])) {
2679 2680 2681 2682
                virQEMUCapsSet(qemuCaps, virQEMUCapsCommandLine[i].flag);
                break;
            }
        }
2683
        virStringListFree(values);
2684 2685 2686 2687
    }

    return 0;
}
2688

2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702
static int
virQEMUCapsProbeQMPMigrationCapabilities(virQEMUCapsPtr qemuCaps,
                                         qemuMonitorPtr mon)
{
    char **caps = NULL;
    int ncaps;

    if ((ncaps = qemuMonitorGetMigrationCapabilities(mon, &caps)) < 0)
        return -1;

    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsMigration),
                                  virQEMUCapsMigration,
                                  ncaps, caps);
2703
    virStringListFreeCount(caps, ncaps);
2704 2705 2706 2707

    return 0;
}

A
Andrea Bolognani 已提交
2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727
/**
 * virQEMUCapsProbeQMPGICCapabilities:
 * @qemuCaps: QEMU binary capabilities
 * @mon: QEMU monitor
 *
 * Use @mon to obtain information about the GIC capabilities for the
 * corresponding QEMU binary, and store them in @qemuCaps.
 *
 * Returns: 0 on success, <0 on failure
 */
static int
virQEMUCapsProbeQMPGICCapabilities(virQEMUCapsPtr qemuCaps,
                                   qemuMonitorPtr mon)
{
    virGICCapability *caps = NULL;
    int ncaps;

    if ((ncaps = qemuMonitorGetGICCapabilities(mon, &caps)) < 0)
        return -1;

2728
    virQEMUCapsSetGICCapabilities(qemuCaps, caps, ncaps);
A
Andrea Bolognani 已提交
2729 2730 2731 2732

    return 0;
}

2733

2734
/* Returns -1 on error, 0 if SEV is not supported, 1 if SEV is supported */
2735 2736 2737 2738
static int
virQEMUCapsProbeQMPSEVCapabilities(virQEMUCapsPtr qemuCaps,
                                   qemuMonitorPtr mon)
{
2739
    int rc = -1;
2740 2741
    virSEVCapability *caps = NULL;

2742 2743
    if ((rc = qemuMonitorGetSEVCapabilities(mon, &caps)) <= 0)
        return rc;
2744

2745 2746
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);
    qemuCaps->sevCapabilities = caps;
2747
    return rc;
2748 2749 2750
}


2751
bool
2752
virQEMUCapsCPUFilterFeatures(const char *name,
2753
                             void *opaque)
2754
{
2755
    virArch *arch = opaque;
2756

2757
    if (!ARCH_IS_X86(*arch))
2758 2759
        return true;

2760 2761 2762 2763 2764 2765 2766 2767 2768
    if (STREQ(name, "cmt") ||
        STREQ(name, "mbm_total") ||
        STREQ(name, "mbm_local"))
        return false;

    return true;
}


2769 2770 2771
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
 *          1 when the caller should fall back to using virCapsPtr->host.cpu,
2772
 *          2 when cpu model info is not supported for this configuration,
2773 2774 2775 2776
 *         -1 on error.
 */
static int
virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps,
2777
                            virDomainVirtType type,
2778
                            qemuMonitorCPUModelInfoPtr modelInfo,
2779 2780
                            virCPUDefPtr cpu,
                            bool migratable)
2781
{
2782
    size_t i;
2783

2784
    if (!modelInfo) {
2785 2786 2787 2788 2789 2790 2791 2792
        if (type == VIR_DOMAIN_VIRT_KVM) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("missing host CPU model info from QEMU "
                             "capabilities for binary %s"),
                           qemuCaps->binary);
            return -1;
        }
        return 2;
2793
    }
J
Jiri Denemark 已提交
2794

2795 2796
    if (VIR_STRDUP(cpu->model, modelInfo->name) < 0 ||
        VIR_ALLOC_N(cpu->features, modelInfo->nprops) < 0)
2797
        return -1;
2798 2799 2800 2801 2802

    cpu->nfeatures_max = modelInfo->nprops;
    cpu->nfeatures = 0;

    for (i = 0; i < modelInfo->nprops; i++) {
2803 2804
        virCPUFeatureDefPtr feature = cpu->features + cpu->nfeatures;
        qemuMonitorCPUPropertyPtr prop = modelInfo->props + i;
2805

2806 2807
        if (prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN)
            continue;
2808

2809 2810
        if (VIR_STRDUP(feature->name, prop->name) < 0)
            return -1;
2811 2812 2813 2814 2815 2816

        if (!prop->value.boolean ||
            (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
            feature->policy = VIR_CPU_FEATURE_DISABLE;
        else
            feature->policy = VIR_CPU_FEATURE_REQUIRE;
2817 2818 2819
        cpu->nfeatures++;
    }

2820 2821
    return 0;
}
2822

2823

2824 2825 2826 2827 2828 2829 2830 2831
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
 *          1 when the caller should fall back to using virCapsPtr->host.cpu,
 *         -1 on error.
 */
static int
virQEMUCapsInitCPUModelX86(virQEMUCapsPtr qemuCaps,
                           virDomainVirtType type,
2832
                           qemuMonitorCPUModelInfoPtr model,
2833 2834
                           virCPUDefPtr cpu,
                           bool migratable)
2835 2836 2837 2838
{
    virCPUDataPtr data = NULL;
    unsigned long long sigFamily = 0;
    unsigned long long sigModel = 0;
2839
    unsigned long long sigStepping = 0;
2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853
    int ret = -1;
    size_t i;

    if (!model)
        return 1;

    if (!(data = virCPUDataNew(VIR_ARCH_X86_64)))
        goto cleanup;

    for (i = 0; i < model->nprops; i++) {
        qemuMonitorCPUPropertyPtr prop = model->props + i;

        switch (prop->type) {
        case QEMU_MONITOR_CPU_PROPERTY_BOOLEAN:
2854 2855 2856 2857 2858
            if (!prop->value.boolean ||
                (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
                continue;

            if (virCPUx86DataAddFeature(data, prop->name) < 0)
2859
                goto cleanup;
2860

2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873
            break;

        case QEMU_MONITOR_CPU_PROPERTY_STRING:
            if (STREQ(prop->name, "vendor") &&
                virCPUx86DataSetVendor(data, prop->value.string) < 0)
                goto cleanup;
            break;

        case QEMU_MONITOR_CPU_PROPERTY_NUMBER:
            if (STREQ(prop->name, "family"))
                sigFamily = prop->value.number;
            else if (STREQ(prop->name, "model"))
                sigModel = prop->value.number;
2874 2875
            else if (STREQ(prop->name, "stepping"))
                sigStepping = prop->value.number;
2876 2877 2878 2879 2880 2881 2882
            break;

        case QEMU_MONITOR_CPU_PROPERTY_LAST:
            break;
        }
    }

2883
    if (virCPUx86DataSetSignature(data, sigFamily, sigModel, sigStepping) < 0)
2884 2885
        goto cleanup;

2886
    if (cpuDecode(cpu, data, virQEMUCapsGetCPUDefinitions(qemuCaps, type)) < 0)
2887 2888 2889 2890 2891 2892 2893 2894 2895 2896
        goto cleanup;

    ret = 0;

 cleanup:
    virCPUDataFree(data);
    return ret;
}


2897 2898
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
2899 2900
 *          1 when the caller should fall back to other methods,
 *          2 when cpu model info is not supported for this configuration,
2901 2902
 *         -1 on error.
 */
2903
int
2904
virQEMUCapsInitCPUModel(virQEMUCapsPtr qemuCaps,
2905
                        virDomainVirtType type,
2906 2907
                        virCPUDefPtr cpu,
                        bool migratable)
2908
{
2909
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
2910 2911
    int ret = 1;

2912
    if (migratable && cpuData->info && !cpuData->info->migratability)
2913 2914
        return 1;

2915
    if (ARCH_IS_S390(qemuCaps->arch)) {
2916
        ret = virQEMUCapsInitCPUModelS390(qemuCaps, type, cpuData->info,
2917 2918 2919 2920 2921
                                          cpu, migratable);
    } else if (ARCH_IS_X86(qemuCaps->arch)) {
        ret = virQEMUCapsInitCPUModelX86(qemuCaps, type, cpuData->info,
                                         cpu, migratable);
    }
2922

2923 2924 2925
    if (ret == 0)
        cpu->fallback = VIR_CPU_FALLBACK_FORBID;

2926
    return ret;
2927 2928 2929
}


2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946
static virCPUDefPtr
virQEMUCapsNewHostCPUModel(void)
{
    virCPUDefPtr cpu;

    if (VIR_ALLOC(cpu) < 0)
        return NULL;

    cpu->type = VIR_CPU_TYPE_GUEST;
    cpu->mode = VIR_CPU_MODE_CUSTOM;
    cpu->match = VIR_CPU_MATCH_EXACT;
    cpu->fallback = VIR_CPU_FALLBACK_ALLOW;

    return cpu;
}


2947 2948
void
virQEMUCapsInitHostCPUModel(virQEMUCapsPtr qemuCaps,
2949
                            virArch hostArch,
2950
                            virDomainVirtType type)
2951 2952
{
    virCPUDefPtr cpu = NULL;
2953
    virCPUDefPtr cpuExpanded = NULL;
2954
    virCPUDefPtr migCPU = NULL;
2955
    virCPUDefPtr hostCPU = NULL;
2956 2957
    virCPUDefPtr fullCPU = NULL;
    size_t i;
2958
    int rc;
2959

2960
    if (!virQEMUCapsGuestIsNative(hostArch, qemuCaps->arch))
2961 2962
        return;

2963
    if (!(cpu = virQEMUCapsNewHostCPUModel()))
2964 2965
        goto error;

2966
    if ((rc = virQEMUCapsInitCPUModel(qemuCaps, type, cpu, false)) < 0) {
2967 2968
        goto error;
    } else if (rc == 1) {
2969
        VIR_DEBUG("No host CPU model info from QEMU; probing host CPU directly");
2970

2971
        hostCPU = virQEMUCapsProbeHostCPUForEmulator(hostArch, qemuCaps, type);
2972 2973
        if (!hostCPU ||
            virCPUDefCopyModelFilter(cpu, hostCPU, true,
2974
                                     virQEMUCapsCPUFilterFeatures,
2975
                                     &qemuCaps->arch) < 0)
2976
            goto error;
2977 2978 2979 2980 2981
    } else if (rc == 2) {
        VIR_DEBUG("QEMU does not provide CPU model for arch=%s virttype=%s",
                  virArchToString(qemuCaps->arch),
                  virDomainVirtTypeToString(type));
        goto error;
2982 2983 2984
    } else if (type == VIR_DOMAIN_VIRT_KVM &&
               virCPUGetHostIsSupported(qemuCaps->arch)) {
        if (!(fullCPU = virCPUGetHost(qemuCaps->arch, VIR_CPU_TYPE_GUEST,
2985
                                      NULL, NULL)))
2986 2987
            goto error;

2988 2989 2990 2991 2992 2993 2994
        if (!(cpuExpanded = virCPUDefCopy(cpu)) ||
            virCPUExpandFeatures(qemuCaps->arch, cpuExpanded) < 0)
            goto error;

        for (i = 0; i < cpuExpanded->nfeatures; i++) {
            if (cpuExpanded->features[i].policy == VIR_CPU_FEATURE_REQUIRE &&
                virCPUDefUpdateFeature(fullCPU, cpuExpanded->features[i].name,
2995 2996 2997
                                       VIR_CPU_FEATURE_REQUIRE) < 0)
                goto error;
        }
2998 2999
    }

3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
    if (!(migCPU = virQEMUCapsNewHostCPUModel()))
        goto error;

    if ((rc = virQEMUCapsInitCPUModel(qemuCaps, type, migCPU, true)) < 0) {
        goto error;
    } else if (rc == 1) {
        VIR_DEBUG("CPU migratability not provided by QEMU");

        virCPUDefFree(migCPU);
        if (!(migCPU = virCPUCopyMigratable(qemuCaps->arch, cpu)))
            goto error;
    }

3013
    virQEMUCapsSetHostModel(qemuCaps, type, cpu, migCPU, fullCPU);
3014

3015
 cleanup:
3016
    virCPUDefFree(cpuExpanded);
3017
    virCPUDefFree(hostCPU);
3018 3019 3020 3021
    return;

 error:
    virCPUDefFree(cpu);
3022
    virCPUDefFree(migCPU);
3023
    virCPUDefFree(fullCPU);
3024
    virResetLastError();
3025
    goto cleanup;
3026 3027 3028
}


3029 3030 3031 3032 3033
void
virQEMUCapsSetCPUModelInfo(virQEMUCapsPtr qemuCaps,
                           virDomainVirtType type,
                           qemuMonitorCPUModelInfoPtr modelInfo)
{
3034 3035 3036
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

    cpuData->info = modelInfo;
3037 3038 3039
}


3040 3041
static int
virQEMUCapsLoadHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3042 3043
                                xmlXPathContextPtr ctxt,
                                virDomainVirtType virtType)
3044 3045 3046
{
    char *str = NULL;
    xmlNodePtr hostCPUNode;
3047
    xmlNodePtr *nodes = NULL;
3048 3049 3050 3051 3052
    xmlNodePtr oldnode = ctxt->node;
    qemuMonitorCPUModelInfoPtr hostCPU = NULL;
    int ret = -1;
    size_t i;
    int n;
3053
    int val;
3054

3055 3056 3057 3058 3059 3060
    if (virtType == VIR_DOMAIN_VIRT_KVM)
        hostCPUNode = virXPathNode("./hostCPU[@type='kvm']", ctxt);
    else
        hostCPUNode = virXPathNode("./hostCPU[@type='tcg']", ctxt);

    if (!hostCPUNode) {
3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074
        ret = 0;
        goto cleanup;
    }

    if (VIR_ALLOC(hostCPU) < 0)
        goto cleanup;

    if (!(hostCPU->name = virXMLPropString(hostCPUNode, "model"))) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing host CPU model name in QEMU "
                         "capabilities cache"));
        goto cleanup;
    }

3075 3076 3077 3078 3079 3080 3081 3082 3083
    if (!(str = virXMLPropString(hostCPUNode, "migratability")) ||
        (val = virTristateBoolTypeFromString(str)) <= 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid migratability value for host CPU model"));
        goto cleanup;
    }
    hostCPU->migratability = val == VIR_TRISTATE_BOOL_YES;
    VIR_FREE(str);

3084 3085
    ctxt->node = hostCPUNode;

3086
    if ((n = virXPathNodeSet("./property", ctxt, &nodes)) > 0) {
3087 3088 3089 3090 3091 3092
        if (VIR_ALLOC_N(hostCPU->props, n) < 0)
            goto cleanup;

        hostCPU->nprops = n;

        for (i = 0; i < n; i++) {
3093 3094 3095 3096 3097
            qemuMonitorCPUPropertyPtr prop = hostCPU->props + i;

            ctxt->node = nodes[i];

            if (!(prop->name = virXMLPropString(ctxt->node, "name"))) {
3098 3099
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing 'name' attribute for a host CPU"
3100
                                 " model property in QEMU capabilities cache"));
3101 3102 3103
                goto cleanup;
            }

3104
            if (!(str = virXMLPropString(ctxt->node, "type")) ||
3105
                (val = qemuMonitorCPUPropertyTypeFromString(str)) < 0) {
3106
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3107 3108
                               _("missing or invalid CPU model property type "
                                 "in QEMU capabilities cache"));
3109 3110 3111
                goto cleanup;
            }
            VIR_FREE(str);
3112

3113
            prop->type = val;
3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144
            switch (prop->type) {
            case QEMU_MONITOR_CPU_PROPERTY_BOOLEAN:
                if (virXPathBoolean("./@value='true'", ctxt))
                    prop->value.boolean = true;
                break;

            case QEMU_MONITOR_CPU_PROPERTY_STRING:
                prop->value.string = virXMLPropString(ctxt->node, "value");
                if (!prop->value.string) {
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("invalid string value for '%s' host CPU "
                                     "model property in QEMU capabilities cache"),
                                   prop->name);
                    goto cleanup;
                }
                break;

            case QEMU_MONITOR_CPU_PROPERTY_NUMBER:
                if (virXPathLongLong("string(./@value)", ctxt,
                                     &prop->value.number) < 0) {
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("invalid number value for '%s' host CPU "
                                     "model property in QEMU capabilities cache"),
                                   prop->name);
                    goto cleanup;
                }
                break;

            case QEMU_MONITOR_CPU_PROPERTY_LAST:
                break;
            }
3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157

            if ((str = virXMLPropString(ctxt->node, "migratable"))) {
                if ((val = virTristateBoolTypeFromString(str)) <= 0) {
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("unknown migratable value for '%s' host "
                                     "CPU model property"),
                                   prop->name);
                    goto cleanup;
                }

                prop->migratable = val;
                VIR_FREE(str);
            }
3158 3159 3160
        }
    }

3161
    virQEMUCapsSetCPUModelInfo(qemuCaps, virtType, hostCPU);
3162 3163 3164 3165 3166 3167
    hostCPU = NULL;
    ret = 0;

 cleanup:
    ctxt->node = oldnode;
    VIR_FREE(str);
3168
    VIR_FREE(nodes);
3169 3170 3171 3172 3173
    qemuMonitorCPUModelInfoFree(hostCPU);
    return ret;
}


3174 3175
static int
virQEMUCapsLoadCPUModels(virQEMUCapsPtr qemuCaps,
3176 3177
                         xmlXPathContextPtr ctxt,
                         virDomainVirtType type)
3178
{
3179
    virDomainCapsCPUModelsPtr cpus = NULL;
3180 3181 3182 3183 3184
    xmlNodePtr *nodes = NULL;
    char *str = NULL;
    size_t i;
    int n;
    int ret = -1;
3185 3186 3187 3188
    xmlNodePtr node;
    xmlNodePtr *blockerNodes = NULL;
    char **blockers = NULL;
    int nblockers;
3189

3190 3191 3192 3193 3194 3195
    if (type == VIR_DOMAIN_VIRT_KVM)
        n = virXPathNodeSet("./cpu[@type='kvm']", ctxt, &nodes);
    else
        n = virXPathNodeSet("./cpu[@type='tcg']", ctxt, &nodes);

    if (n < 0) {
3196 3197 3198 3199 3200 3201 3202 3203 3204 3205
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities cpus"));
        goto cleanup;
    }

    if (n == 0) {
        ret = 0;
        goto cleanup;
    }

3206
    if (!(cpus = virDomainCapsCPUModelsNew(n)))
3207 3208
        goto cleanup;

3209 3210 3211 3212 3213
    if (type == VIR_DOMAIN_VIRT_KVM)
        qemuCaps->kvmCPUModels = cpus;
    else
        qemuCaps->tcgCPUModels = cpus;

3214
    for (i = 0; i < n; i++) {
3215 3216 3217 3218 3219 3220 3221 3222 3223 3224
        int usable = VIR_DOMCAPS_CPU_USABLE_UNKNOWN;

        if ((str = virXMLPropString(nodes[i], "usable")) &&
            (usable = virDomainCapsCPUUsableTypeFromString(str)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown value '%s' in attribute 'usable'"), str);
            goto cleanup;
        }
        VIR_FREE(str);

3225 3226 3227 3228 3229 3230
        if (!(str = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("missing cpu name in QEMU capabilities cache"));
            goto cleanup;
        }

3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255
        node = ctxt->node;
        ctxt->node = nodes[i];
        nblockers = virXPathNodeSet("./blocker", ctxt, &blockerNodes);
        ctxt->node = node;

        if (nblockers < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to parse CPU blockers in QEMU capabilities"));
            goto cleanup;
        }

        if (nblockers > 0) {
            size_t j;

            if (VIR_ALLOC_N(blockers, nblockers + 1) < 0)
                goto cleanup;

            for (j = 0; j < nblockers; j++) {
                if (!(blockers[j] = virXMLPropString(blockerNodes[j], "name"))) {
                    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                   _("missing blocker name in QEMU "
                                     "capabilities cache"));
                    goto cleanup;
                }
            }
3256
            VIR_FREE(blockerNodes);
3257 3258 3259
        }

        if (virDomainCapsCPUModelsAddSteal(cpus, &str, usable, &blockers) < 0)
3260 3261 3262 3263 3264 3265 3266 3267
            goto cleanup;
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
    VIR_FREE(str);
3268 3269
    VIR_FREE(blockerNodes);
    virStringListFree(blockers);
3270 3271 3272 3273
    return ret;
}


3274 3275 3276 3277 3278
struct _virQEMUCapsCachePriv {
    char *libDir;
    uid_t runUid;
    gid_t runGid;
    virArch hostArch;
3279
    unsigned int microcodeVersion;
3280
    char *kernelVersion;
3281 3282 3283 3284

    /* cache whether /dev/kvm is usable as runUid:runGuid */
    virTristateBool kvmUsable;
    time_t kvmCtime;
3285 3286 3287 3288 3289
};
typedef struct _virQEMUCapsCachePriv virQEMUCapsCachePriv;
typedef virQEMUCapsCachePriv *virQEMUCapsCachePrivPtr;


3290
static void
3291
virQEMUCapsCachePrivFree(void *privData)
3292
{
3293 3294
    virQEMUCapsCachePrivPtr priv = privData;

3295
    VIR_FREE(priv->libDir);
3296
    VIR_FREE(priv->kernelVersion);
3297 3298 3299 3300
    VIR_FREE(priv);
}


3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352
static int
virQEMUCapsParseSEVInfo(virQEMUCapsPtr qemuCaps, xmlXPathContextPtr ctxt)
{
    VIR_AUTOPTR(virSEVCapability) sev = NULL;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST))
        return 0;

    if (virXPathBoolean("boolean(./sev)", ctxt) == 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing SEV platform data in QEMU "
                         "capabilities cache"));
        return -1;
    }

    if (VIR_ALLOC(sev) < 0)
        return -1;

    if (virXPathUInt("string(./sev/cbitpos)", ctxt, &sev->cbitpos) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing or malformed SEV cbitpos information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    if (virXPathUInt("string(./sev/reducedPhysBits)", ctxt,
                     &sev->reduced_phys_bits) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing or malformed SEV reducedPhysBits information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    if (!(sev->pdh = virXPathString("string(./sev/pdh)", ctxt)))  {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing SEV pdh information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    if (!(sev->cert_chain = virXPathString("string(./sev/certChain)", ctxt))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing SEV certChain information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    VIR_STEAL_PTR(qemuCaps->sevCapabilities, sev);
    return 0;
}


3353 3354 3355 3356 3357 3358
/*
 * Parsing a doc that looks like
 *
 * <qemuCaps>
 *   <qemuctime>234235253</qemuctime>
 *   <selfctime>234235253</selfctime>
3359
 *   <selfvers>1002016</selfvers>
3360 3361 3362 3363 3364 3365
 *   <usedQMP/>
 *   <flag name='foo'/>
 *   <flag name='bar'/>
 *   ...
 *   <cpu name="pentium3"/>
 *   ...
3366
 *   <machine name="pc-1.0" alias="pc" hotplugCpus='yes' maxCpus="4" default="yes"/>
3367 3368 3369
 *   ...
 * </qemuCaps>
 */
3370
int
3371
virQEMUCapsLoadCache(virArch hostArch,
3372
                     virQEMUCapsPtr qemuCaps,
3373
                     const char *filename)
3374 3375 3376 3377 3378 3379 3380
{
    xmlDocPtr doc = NULL;
    int ret = -1;
    size_t i;
    int n;
    xmlNodePtr *nodes = NULL;
    xmlXPathContextPtr ctxt = NULL;
J
Ján Tomko 已提交
3381
    char *str = NULL;
3382
    long long int l;
3383
    unsigned long lu;
3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407

    if (!(doc = virXMLParseFile(filename)))
        goto cleanup;

    if (!(ctxt = xmlXPathNewContext(doc))) {
        virReportOOMError();
        goto cleanup;
    }

    ctxt->node = xmlDocGetRootElement(doc);

    if (STRNEQ((const char *)ctxt->node->name, "qemuCaps")) {
        virReportError(VIR_ERR_XML_ERROR,
                       _("unexpected root element <%s>, "
                         "expecting <qemuCaps>"),
                       ctxt->node->name);
        goto cleanup;
    }

    if (virXPathLongLong("string(./qemuctime)", ctxt, &l) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing qemuctime in QEMU capabilities XML"));
        goto cleanup;
    }
3408
    qemuCaps->ctime = (time_t)l;
3409 3410 3411 3412 3413 3414

    if (virXPathLongLong("string(./selfctime)", ctxt, &l) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing selfctime in QEMU capabilities XML"));
        goto cleanup;
    }
3415
    qemuCaps->libvirtCtime = (time_t)l;
3416

3417
    qemuCaps->libvirtVersion = 0;
3418
    if (virXPathULong("string(./selfvers)", ctxt, &lu) == 0)
3419
        qemuCaps->libvirtVersion = lu;
3420

3421 3422 3423 3424 3425 3426 3427 3428 3429
    qemuCaps->usedQMP = virXPathBoolean("count(./usedQMP) > 0",
                                        ctxt) > 0;

    if ((n = virXPathNodeSet("./flag", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities flags"));
        goto cleanup;
    }
    VIR_DEBUG("Got flags %d", n);
3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441
    for (i = 0; i < n; i++) {
        int flag;
        if (!(str = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("missing flag name in QEMU capabilities cache"));
            goto cleanup;
        }
        flag = virQEMUCapsTypeFromString(str);
        if (flag < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown qemu capabilities flag %s"), str);
            goto cleanup;
3442
        }
3443 3444
        VIR_FREE(str);
        virQEMUCapsSet(qemuCaps, flag);
3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459
    }
    VIR_FREE(nodes);

    if (virXPathUInt("string(./version)", ctxt, &qemuCaps->version) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing version in QEMU capabilities cache"));
        goto cleanup;
    }

    if (virXPathUInt("string(./kvmVersion)", ctxt, &qemuCaps->kvmVersion) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing version in QEMU capabilities cache"));
        goto cleanup;
    }

3460 3461 3462 3463 3464 3465 3466
    if (virXPathUInt("string(./microcodeVersion)", ctxt,
                     &qemuCaps->microcodeVersion) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing microcode version in QEMU capabilities cache"));
        goto cleanup;
    }

3467 3468 3469 3470 3471 3472
    if (virXPathBoolean("boolean(./package)", ctxt) > 0) {
        qemuCaps->package = virXPathString("string(./package)", ctxt);
        if (!qemuCaps->package &&
            VIR_STRDUP(qemuCaps->package, "") < 0)
            goto cleanup;
    }
3473

3474 3475 3476 3477 3478 3479
    if (virXPathBoolean("boolean(./kernelVersion)", ctxt) > 0) {
        qemuCaps->kernelVersion = virXPathString("string(./kernelVersion)", ctxt);
        if (!qemuCaps->kernelVersion)
            goto cleanup;
    }

3480 3481 3482 3483 3484 3485 3486 3487 3488 3489
    if (!(str = virXPathString("string(./arch)", ctxt))) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing arch in QEMU capabilities cache"));
        goto cleanup;
    }
    if (!(qemuCaps->arch = virArchFromString(str))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown arch %s in QEMU capabilities cache"), str);
        goto cleanup;
    }
J
Ján Tomko 已提交
3490
    VIR_FREE(str);
3491

3492 3493
    if (virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3494 3495
        goto cleanup;

3496 3497
    if (virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3498 3499 3500 3501 3502 3503 3504 3505 3506
        goto cleanup;

    if ((n = virXPathNodeSet("./machine", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities machines"));
        goto cleanup;
    }
    if (n > 0) {
        qemuCaps->nmachineTypes = n;
3507
        if (VIR_ALLOC_N(qemuCaps->machineTypes, qemuCaps->nmachineTypes) < 0)
3508 3509 3510
            goto cleanup;

        for (i = 0; i < n; i++) {
3511
            if (!(qemuCaps->machineTypes[i].name = virXMLPropString(nodes[i], "name"))) {
3512 3513 3514 3515
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing machine name in QEMU capabilities cache"));
                goto cleanup;
            }
3516
            qemuCaps->machineTypes[i].alias = virXMLPropString(nodes[i], "alias");
3517 3518 3519

            str = virXMLPropString(nodes[i], "maxCpus");
            if (str &&
3520
                virStrToLong_ui(str, NULL, 10, &(qemuCaps->machineTypes[i].maxCpus)) < 0) {
3521 3522 3523 3524
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed machine cpu count in QEMU capabilities cache"));
                goto cleanup;
            }
J
Ján Tomko 已提交
3525
            VIR_FREE(str);
3526 3527 3528 3529 3530

            str = virXMLPropString(nodes[i], "hotplugCpus");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].hotplugCpus = true;
            VIR_FREE(str);
3531 3532 3533 3534 3535

            str = virXMLPropString(nodes[i], "default");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].qemuDefault = true;
            VIR_FREE(str);
3536 3537 3538 3539
        }
    }
    VIR_FREE(nodes);

A
Andrea Bolognani 已提交
3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605
    if ((n = virXPathNodeSet("./gic", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities gic"));
        goto cleanup;
    }
    if (n > 0) {
        unsigned int uintValue;
        bool boolValue;

        qemuCaps->ngicCapabilities = n;
        if (VIR_ALLOC_N(qemuCaps->gicCapabilities, n) < 0)
            goto cleanup;

        for (i = 0; i < n; i++) {
            virGICCapabilityPtr cap = &qemuCaps->gicCapabilities[i];

            if (!(str = virXMLPropString(nodes[i], "version"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing GIC version "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (virStrToLong_ui(str, NULL, 10, &uintValue) < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed GIC version "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            cap->version = uintValue;
            VIR_FREE(str);

            if (!(str = virXMLPropString(nodes[i], "kernel"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing in-kernel GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (!(boolValue = STREQ(str, "yes")) && STRNEQ(str, "no")) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed in-kernel GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (boolValue)
                cap->implementation |= VIR_GIC_IMPLEMENTATION_KERNEL;
            VIR_FREE(str);

            if (!(str = virXMLPropString(nodes[i], "emulated"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing emulated GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (!(boolValue = STREQ(str, "yes")) && STRNEQ(str, "no")) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed emulated GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (boolValue)
                cap->implementation |= VIR_GIC_IMPLEMENTATION_EMULATED;
            VIR_FREE(str);
        }
    }
    VIR_FREE(nodes);

3606 3607 3608
    if (virQEMUCapsParseSEVInfo(qemuCaps, ctxt) < 0)
        goto cleanup;

3609 3610
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
3611

3612 3613 3614
    if (virXPathBoolean("boolean(./kvmSupportsNesting)", ctxt) > 0)
        qemuCaps->kvmSupportsNesting = true;

3615
    ret = 0;
3616
 cleanup:
J
Ján Tomko 已提交
3617
    VIR_FREE(str);
3618 3619 3620 3621 3622 3623 3624
    VIR_FREE(nodes);
    xmlXPathFreeContext(ctxt);
    xmlFreeDoc(doc);
    return ret;
}


3625 3626
static void
virQEMUCapsFormatHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3627 3628
                                  virBufferPtr buf,
                                  virDomainVirtType type)
3629
{
3630 3631 3632
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
    qemuMonitorCPUModelInfoPtr model = cpuData->info;
    const char *typeStr = type == VIR_DOMAIN_VIRT_KVM ? "kvm" : "tcg";
3633 3634
    size_t i;

3635 3636 3637
    if (!model)
        return;

3638 3639 3640 3641
    virBufferAsprintf(buf,
                      "<hostCPU type='%s' model='%s' migratability='%s'>\n",
                      typeStr, model->name,
                      model->migratability ? "yes" : "no");
3642 3643 3644
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < model->nprops; i++) {
3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667
        qemuMonitorCPUPropertyPtr prop = model->props + i;

        virBufferAsprintf(buf, "<property name='%s' type='%s' ",
                          prop->name,
                          qemuMonitorCPUPropertyTypeToString(prop->type));

        switch (prop->type) {
        case QEMU_MONITOR_CPU_PROPERTY_BOOLEAN:
            virBufferAsprintf(buf, "value='%s'",
                              prop->value.boolean ? "true" : "false");
            break;

        case QEMU_MONITOR_CPU_PROPERTY_STRING:
            virBufferEscapeString(buf, "value='%s'", prop->value.string);
            break;

        case QEMU_MONITOR_CPU_PROPERTY_NUMBER:
            virBufferAsprintf(buf, "value='%lld'", prop->value.number);
            break;

        case QEMU_MONITOR_CPU_PROPERTY_LAST:
            break;
        }
3668 3669 3670 3671 3672

        if (prop->migratable > 0)
            virBufferAsprintf(buf, " migratable='%s'",
                              virTristateBoolTypeToString(prop->migratable));

3673
        virBufferAddLit(buf, "/>\n");
3674 3675 3676 3677 3678 3679 3680
    }

    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</hostCPU>\n");
}


3681 3682
static void
virQEMUCapsFormatCPUModels(virQEMUCapsPtr qemuCaps,
3683 3684
                           virBufferPtr buf,
                           virDomainVirtType type)
3685
{
3686 3687
    virDomainCapsCPUModelsPtr cpus;
    const char *typeStr;
3688 3689
    size_t i;

3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701
    if (type == VIR_DOMAIN_VIRT_KVM) {
        typeStr = "kvm";
        cpus = qemuCaps->kvmCPUModels;
    } else {
        typeStr = "tcg";
        cpus = qemuCaps->tcgCPUModels;
    }

    if (!cpus)
        return;

    for (i = 0; i < cpus->nmodels; i++) {
3702 3703
        virDomainCapsCPUModelPtr cpu = cpus->models + i;

3704
        virBufferAsprintf(buf, "<cpu type='%s' ", typeStr);
3705 3706 3707 3708 3709
        virBufferEscapeString(buf, "name='%s'", cpu->name);
        if (cpu->usable) {
            virBufferAsprintf(buf, " usable='%s'",
                              virDomainCapsCPUUsableTypeToString(cpu->usable));
        }
3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724

        if (cpu->blockers) {
            size_t j;

            virBufferAddLit(buf, ">\n");
            virBufferAdjustIndent(buf, 2);

            for (j = 0; cpu->blockers[j]; j++)
                virBufferAsprintf(buf, "<blocker name='%s'/>\n", cpu->blockers[j]);

            virBufferAdjustIndent(buf, -2);
            virBufferAddLit(buf, "</cpu>\n");
        } else {
            virBufferAddLit(buf, "/>\n");
        }
3725 3726 3727 3728
    }
}


3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746
static void
virQEMUCapsFormatSEVInfo(virQEMUCapsPtr qemuCaps, virBufferPtr buf)
{
    virSEVCapabilityPtr sev = virQEMUCapsGetSEVCapabilities(qemuCaps);

    virBufferAddLit(buf, "<sev>\n");
    virBufferAdjustIndent(buf, 2);
    virBufferAsprintf(buf, "<cbitpos>%u</cbitpos>\n", sev->cbitpos);
    virBufferAsprintf(buf, "<reducedPhysBits>%u</reducedPhysBits>\n",
                      sev->reduced_phys_bits);
    virBufferEscapeString(buf, "<pdh>%s</pdh>\n", sev->pdh);
    virBufferEscapeString(buf, "<certChain>%s</certChain>\n",
                          sev->cert_chain);
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</sev>\n");
}


3747
char *
3748
virQEMUCapsFormatCache(virQEMUCapsPtr qemuCaps)
3749 3750
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
3751
    char *ret = NULL;
3752 3753 3754
    size_t i;

    virBufferAddLit(&buf, "<qemuCaps>\n");
3755
    virBufferAdjustIndent(&buf, 2);
3756

3757
    virBufferAsprintf(&buf, "<qemuctime>%llu</qemuctime>\n",
3758
                      (long long)qemuCaps->ctime);
3759
    virBufferAsprintf(&buf, "<selfctime>%llu</selfctime>\n",
3760
                      (long long)qemuCaps->libvirtCtime);
3761
    virBufferAsprintf(&buf, "<selfvers>%lu</selfvers>\n",
3762
                      (unsigned long)qemuCaps->libvirtVersion);
3763 3764

    if (qemuCaps->usedQMP)
3765
        virBufferAddLit(&buf, "<usedQMP/>\n");
3766 3767 3768

    for (i = 0; i < QEMU_CAPS_LAST; i++) {
        if (virQEMUCapsGet(qemuCaps, i)) {
3769
            virBufferAsprintf(&buf, "<flag name='%s'/>\n",
3770 3771 3772 3773
                              virQEMUCapsTypeToString(i));
        }
    }

3774
    virBufferAsprintf(&buf, "<version>%d</version>\n",
3775 3776
                      qemuCaps->version);

3777
    virBufferAsprintf(&buf, "<kvmVersion>%d</kvmVersion>\n",
3778 3779
                      qemuCaps->kvmVersion);

3780 3781 3782
    virBufferAsprintf(&buf, "<microcodeVersion>%u</microcodeVersion>\n",
                      qemuCaps->microcodeVersion);

3783 3784 3785 3786
    if (qemuCaps->package)
        virBufferAsprintf(&buf, "<package>%s</package>\n",
                          qemuCaps->package);

3787 3788 3789 3790
    if (qemuCaps->kernelVersion)
        virBufferAsprintf(&buf, "<kernelVersion>%s</kernelVersion>\n",
                          qemuCaps->kernelVersion);

3791
    virBufferAsprintf(&buf, "<arch>%s</arch>\n",
3792 3793
                      virArchToString(qemuCaps->arch));

3794 3795
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3796

3797 3798
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3799 3800

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
3801
        virBufferEscapeString(&buf, "<machine name='%s'",
3802 3803
                              qemuCaps->machineTypes[i].name);
        if (qemuCaps->machineTypes[i].alias)
3804
            virBufferEscapeString(&buf, " alias='%s'",
3805
                              qemuCaps->machineTypes[i].alias);
3806 3807
        if (qemuCaps->machineTypes[i].hotplugCpus)
            virBufferAddLit(&buf, " hotplugCpus='yes'");
3808
        virBufferAsprintf(&buf, " maxCpus='%u'",
3809
                          qemuCaps->machineTypes[i].maxCpus);
3810 3811 3812
        if (qemuCaps->machineTypes[i].qemuDefault)
            virBufferAddLit(&buf, " default='yes'");
        virBufferAddLit(&buf, "/>\n");
3813 3814
    }

A
Andrea Bolognani 已提交
3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830
    for (i = 0; i < qemuCaps->ngicCapabilities; i++) {
        virGICCapabilityPtr cap;
        bool kernel;
        bool emulated;

        cap = &qemuCaps->gicCapabilities[i];
        kernel = (cap->implementation & VIR_GIC_IMPLEMENTATION_KERNEL);
        emulated = (cap->implementation & VIR_GIC_IMPLEMENTATION_EMULATED);

        virBufferAsprintf(&buf,
                          "<gic version='%d' kernel='%s' emulated='%s'/>\n",
                          cap->version,
                          kernel ? "yes" : "no",
                          emulated ? "yes" : "no");
    }

3831 3832 3833
    if (qemuCaps->sevCapabilities)
        virQEMUCapsFormatSEVInfo(qemuCaps, &buf);

3834 3835 3836
    if (qemuCaps->kvmSupportsNesting)
        virBufferAddLit(&buf, "<kvmSupportsNesting/>\n");

3837
    virBufferAdjustIndent(&buf, -2);
3838 3839
    virBufferAddLit(&buf, "</qemuCaps>\n");

3840 3841 3842 3843 3844 3845 3846 3847
    if (virBufferCheckError(&buf) == 0)
        ret = virBufferContentAndReset(&buf);

    return ret;
}


static int
3848 3849 3850
virQEMUCapsSaveFile(void *data,
                    const char *filename,
                    void *privData ATTRIBUTE_UNUSED)
3851
{
3852
    virQEMUCapsPtr qemuCaps = data;
3853 3854
    char *xml = NULL;
    int ret = -1;
3855

3856
    xml = virQEMUCapsFormatCache(qemuCaps);
3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867

    if (virFileWriteStr(filename, xml, 0600) < 0) {
        virReportSystemError(errno,
                             _("Failed to save '%s' for '%s'"),
                             filename, qemuCaps->binary);
        goto cleanup;
    }

    VIR_DEBUG("Saved caps '%s' for '%s' with (%lld, %lld)",
              filename, qemuCaps->binary,
              (long long)qemuCaps->ctime,
3868
              (long long)qemuCaps->libvirtCtime);
3869 3870 3871 3872 3873 3874 3875 3876

    ret = 0;
 cleanup:
    VIR_FREE(xml);
    return ret;
}


3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911
/* Check the kernel module parameters 'nested' file to determine if enabled
 *
 *   Intel: 'kvm_intel' uses 'Y'
 *   AMD:   'kvm_amd' uses '1'
 *   PPC64: 'kvm_hv' uses 'Y'
 *   S390:  'kvm' uses '1'
 */
static bool
virQEMUCapsKVMSupportsNesting(void)
{
    static char const * const kmod[] = {"kvm_intel", "kvm_amd",
                                        "kvm_hv", "kvm"};
    VIR_AUTOFREE(char *) value = NULL;
    int rc;
    size_t i;

    for (i = 0; i < ARRAY_CARDINALITY(kmod); i++) {
        VIR_FREE(value);
        rc = virFileReadValueString(&value, "/sys/module/%s/parameters/nested",
                                    kmod[i]);
        if (rc == -2)
            continue;
        if (rc < 0) {
            virResetLastError();
            return false;
        }

        if (value[0] == 'Y' || value[0] == 'y' || value[0] == '1')
            return true;
    }

    return false;
}


3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959
/* Determine whether '/dev/kvm' is usable as QEMU user:QEMU group. */
static bool
virQEMUCapsKVMUsable(virQEMUCapsCachePrivPtr priv)
{
    struct stat sb;
    static const char *kvm_device = "/dev/kvm";
    virTristateBool value;
    virTristateBool cached_value = priv->kvmUsable;
    time_t kvm_ctime;
    time_t cached_kvm_ctime = priv->kvmCtime;

    if (stat(kvm_device, &sb) < 0) {
        if (errno != ENOENT) {
            virReportSystemError(errno,
                                 _("Failed to stat %s"), kvm_device);
        }
        return false;
    }
    kvm_ctime = sb.st_ctime;

    if (kvm_ctime != cached_kvm_ctime) {
        VIR_DEBUG("%s has changed (%lld vs %lld)", kvm_device,
                  (long long)kvm_ctime, (long long)cached_kvm_ctime);
        cached_value = VIR_TRISTATE_BOOL_ABSENT;
    }

    if (cached_value != VIR_TRISTATE_BOOL_ABSENT)
        return cached_value == VIR_TRISTATE_BOOL_YES;

    if (virFileAccessibleAs(kvm_device, R_OK | W_OK,
                            priv->runUid, priv->runGid) == 0) {
        value = VIR_TRISTATE_BOOL_YES;
    } else {
        value = VIR_TRISTATE_BOOL_NO;
    }

    /* There is a race window between 'stat' and
     * 'virFileAccessibleAs'. However, since we're only interested in
     * detecting changes *after* the virFileAccessibleAs check, we can
     * neglect this here.
     */
    priv->kvmCtime = kvm_ctime;
    priv->kvmUsable = value;

    return value == VIR_TRISTATE_BOOL_YES;
}


3960
static bool
3961 3962
virQEMUCapsIsValid(void *data,
                   void *privData)
3963
{
3964 3965
    virQEMUCapsPtr qemuCaps = data;
    virQEMUCapsCachePrivPtr priv = privData;
3966
    bool kvmUsable;
3967
    struct stat sb;
3968
    bool kvmSupportsNesting;
3969 3970 3971 3972

    if (!qemuCaps->binary)
        return true;

3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984
    if (qemuCaps->libvirtCtime != virGetSelfLastChanged() ||
        qemuCaps->libvirtVersion != LIBVIR_VERSION_NUMBER) {
        VIR_DEBUG("Outdated capabilities for '%s': libvirt changed "
                  "(%lld vs %lld, %lu vs %lu)",
                  qemuCaps->binary,
                  (long long)qemuCaps->libvirtCtime,
                  (long long)virGetSelfLastChanged(),
                  (unsigned long)qemuCaps->libvirtVersion,
                  (unsigned long)LIBVIR_VERSION_NUMBER);
        return false;
    }

3985 3986 3987 3988 3989 3990
    if (stat(qemuCaps->binary, &sb) < 0) {
        char ebuf[1024];
        VIR_DEBUG("Failed to stat QEMU binary '%s': %s",
                  qemuCaps->binary,
                  virStrerror(errno, ebuf, sizeof(ebuf)));
        return false;
3991 3992
    }

3993
    if (sb.st_ctime != qemuCaps->ctime) {
3994 3995 3996
        VIR_DEBUG("Outdated capabilities for '%s': QEMU binary changed "
                  "(%lld vs %lld)",
                  qemuCaps->binary,
3997
                  (long long)sb.st_ctime, (long long)qemuCaps->ctime);
3998 3999 4000
        return false;
    }

4001 4002 4003 4004 4005 4006 4007 4008
    if (!virQEMUCapsGuestIsNative(priv->hostArch, qemuCaps->arch)) {
        VIR_DEBUG("Guest arch (%s) is not native to host arch (%s), "
                  "skipping KVM-related checks",
                  virArchToString(qemuCaps->arch),
                  virArchToString(priv->hostArch));
        return true;
    }

4009
    kvmUsable = virQEMUCapsKVMUsable(priv);
4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM) &&
        kvmUsable) {
        VIR_DEBUG("KVM was not enabled when probing '%s', "
                  "but it should be usable now",
                  qemuCaps->binary);
        return false;
    }

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM) &&
        !kvmUsable) {
        VIR_DEBUG("KVM was enabled when probing '%s', "
                  "but it is not available now",
                  qemuCaps->binary);
        return false;
    }

4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
        if (priv->microcodeVersion != qemuCaps->microcodeVersion) {
            VIR_DEBUG("Outdated capabilities for '%s': microcode version "
                      "changed (%u vs %u)",
                      qemuCaps->binary,
                      priv->microcodeVersion,
                      qemuCaps->microcodeVersion);
            return false;
        }

        if (STRNEQ_NULLABLE(priv->kernelVersion, qemuCaps->kernelVersion)) {
            VIR_DEBUG("Outdated capabilities for '%s': kernel version changed "
                      "('%s' vs '%s')",
                      qemuCaps->binary,
                      priv->kernelVersion,
                      qemuCaps->kernelVersion);
            return false;
        }
4045 4046 4047 4048 4049 4050 4051 4052

        kvmSupportsNesting = virQEMUCapsKVMSupportsNesting();
        if (kvmSupportsNesting != qemuCaps->kvmSupportsNesting) {
            VIR_DEBUG("Outdated capabilities for '%s': kvm kernel nested "
                      "value changed from %d",
                     qemuCaps->binary, qemuCaps->kvmSupportsNesting);
            return false;
        }
4053 4054
    }

4055 4056 4057 4058
    return true;
}


4059
static void virQEMUCapsMonitorNotify(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
4060 4061
                                     virDomainObjPtr vm ATTRIBUTE_UNUSED,
                                     void *opaque ATTRIBUTE_UNUSED)
4062 4063 4064 4065
{
}

static qemuMonitorCallbacks callbacks = {
4066 4067
    .eofNotify = virQEMUCapsMonitorNotify,
    .errorNotify = virQEMUCapsMonitorNotify,
4068 4069 4070
};


4071 4072 4073 4074 4075 4076 4077 4078
/**
 * virQEMUCapsInitQMPArch:
 * @qemuCaps: QEMU capabilities
 * @mon: QEMU monitor
 *
 * Initialize the architecture for @qemuCaps by asking @mon.
 *
 * Returns: 0 on success, <0 on failure
4079 4080
 */
static int
4081
virQEMUCapsInitQMPArch(virQEMUCapsPtr qemuCaps,
4082 4083 4084 4085 4086 4087
                            qemuMonitorPtr mon)
{
    char *archstr = NULL;
    int ret = -1;

    if (!(archstr = qemuMonitorGetTargetArch(mon)))
4088
        goto cleanup;
4089 4090 4091 4092 4093 4094 4095

    if ((qemuCaps->arch = virQEMUCapsArchFromString(archstr)) == VIR_ARCH_NONE) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown QEMU arch %s"), archstr);
        goto cleanup;
    }

4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109
    ret = 0;

 cleanup:
    VIR_FREE(archstr);
    return ret;
}


/**
 * virQEMUCapsInitQMPBasicArch:
 * @qemuCaps: QEMU capabilities
 *
 * Initialize @qemuCaps with basic architecture-dependent capabilities.
 */
4110
void
4111 4112
virQEMUCapsInitQMPBasicArch(virQEMUCapsPtr qemuCaps)
{
4113 4114 4115
    /* ACPI only works on x86 and aarch64 */
    if (ARCH_IS_X86(qemuCaps->arch) ||
        qemuCaps->arch == VIR_ARCH_AARCH64) {
4116
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_ACPI);
4117 4118
    }

J
Ján Tomko 已提交
4119 4120
    /* HPET is x86 specific */
    if (ARCH_IS_X86(qemuCaps->arch))
J
Ján Tomko 已提交
4121
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_HPET);
4122
}
4123

4124 4125 4126 4127 4128 4129

static int
virQEMUCapsProbeQMPSchemaCapabilities(virQEMUCapsPtr qemuCaps,
                                      qemuMonitorPtr mon)
{
    struct virQEMUCapsStringFlags *entry;
4130 4131
    virJSONValuePtr schemareply;
    virHashTablePtr schema = NULL;
4132 4133
    size_t i;

4134
    if (!(schemareply = qemuMonitorQueryQMPSchema(mon)))
4135 4136
        return -1;

4137 4138 4139 4140
    if (!(schema = virQEMUQAPISchemaConvert(schemareply)))
        return -1;
    schemareply = NULL;

4141 4142 4143
    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsQMPSchemaQueries); i++) {
        entry = virQEMUCapsQMPSchemaQueries + i;

4144
        if (virQEMUQAPISchemaPathExists(entry->value, schema))
4145 4146 4147
            virQEMUCapsSet(qemuCaps, entry->flag);
    }

4148 4149 4150 4151 4152 4153 4154 4155
    /* probe also for basic event support */
    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsEvents); i++) {
        entry = virQEMUCapsEvents + i;

        if (virQEMUQAPISchemaPathExists(entry->value, schema))
            virQEMUCapsSet(qemuCaps, entry->flag);
    }

4156 4157 4158 4159
    virHashFree(schema);
    return 0;
}

J
Ján Tomko 已提交
4160
#define QEMU_MIN_MAJOR 1
J
Ján Tomko 已提交
4161
#define QEMU_MIN_MINOR 5
J
Ján Tomko 已提交
4162
#define QEMU_MIN_MICRO 0
4163

4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175
int
virQEMUCapsInitQMPMonitor(virQEMUCapsPtr qemuCaps,
                          qemuMonitorPtr mon)
{
    int ret = -1;
    int major, minor, micro;
    char *package = NULL;

    /* @mon is supposed to be locked by callee */

    if (qemuMonitorSetCapabilities(mon) < 0) {
        VIR_DEBUG("Failed to set monitor capabilities %s",
4176
                  virGetLastErrorMessage());
4177 4178 4179 4180 4181 4182 4183 4184
        ret = 0;
        goto cleanup;
    }

    if (qemuMonitorGetVersion(mon,
                              &major, &minor, &micro,
                              &package) < 0) {
        VIR_DEBUG("Failed to query monitor version %s",
4185
                  virGetLastErrorMessage());
4186 4187 4188 4189 4190 4191 4192
        ret = 0;
        goto cleanup;
    }

    VIR_DEBUG("Got version %d.%d.%d (%s)",
              major, minor, micro, NULLSTR(package));

J
Ján Tomko 已提交
4193 4194 4195 4196 4197 4198
    if (major < QEMU_MIN_MAJOR ||
        (major == QEMU_MIN_MAJOR && minor < QEMU_MIN_MINOR)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("QEMU version >= %d.%d.%d is required, but %d.%d.%d found"),
                       QEMU_MIN_MAJOR, QEMU_MIN_MINOR, QEMU_MIN_MICRO,
                       major, minor, micro);
4199 4200 4201 4202
        goto cleanup;
    }

    qemuCaps->version = major * 1000000 + minor * 1000 + micro;
4203
    qemuCaps->package = package;
4204 4205
    qemuCaps->usedQMP = true;

4206
    if (virQEMUCapsInitQMPArch(qemuCaps, mon) < 0)
4207 4208
        goto cleanup;

4209 4210
    virQEMUCapsInitQMPBasicArch(qemuCaps);

4211 4212 4213
    if (qemuCaps->version >= 1006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_DEVICE_VIDEO_PRIMARY);

4214 4215 4216 4217
    /* vmport option is supported v2.2.0 onwards */
    if (qemuCaps->version >= 2002000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT);

4218 4219 4220 4221 4222 4223
    /* -cpu ...,aarch64=off supported in v2.3.0 and onwards. But it
       isn't detectable via qmp at this point */
    if (qemuCaps->arch == VIR_ARCH_AARCH64 &&
        qemuCaps->version >= 2003000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_CPU_AARCH64_OFF);

4224 4225 4226
    /* vhost-user supports multi-queue from v2.4.0 onwards,
     * but there is no way to query for that capability */
    if (qemuCaps->version >= 2004000)
J
Ján Tomko 已提交
4227
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE);
4228

M
Michal Privoznik 已提交
4229 4230 4231 4232
    /* smm option is supported from v2.4.0 */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT);

4233 4234 4235 4236
    /* sdl -gl option is supported from v2.4.0 (qemu commit id 0b71a5d5) */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_SDL_GL);

4237 4238 4239 4240
    /* Since 2.4.50 ARM virt machine supports gic-version option */
    if (qemuCaps->version >= 2004050)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACH_VIRT_GIC_VERSION);

4241 4242 4243 4244
    /* no way to query if -machine kernel_irqchip supports split */
    if (qemuCaps->version >= 2006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_KERNEL_IRQCHIP_SPLIT);

4245 4246 4247 4248 4249 4250 4251
    /* HPT resizing is supported since QEMU 2.10 on ppc64; unfortunately
     * there's no sane way to probe for it */
    if (qemuCaps->version >= 2010000 &&
        ARCH_IS_PPC64(qemuCaps->arch)) {
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_PSERIES_RESIZE_HPT);
    }

4252 4253 4254 4255 4256
    /* '-display egl-headless' cmdline option is supported since QEMU 2.10, but
     * there's no way to probe it */
    if (qemuCaps->version >= 2010000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_EGL_HEADLESS);

4257 4258 4259 4260
    /* no way to query for -numa dist */
    if (qemuCaps->version >= 2010000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NUMA_DIST);

4261 4262 4263 4264 4265 4266
    /* no way to query max-cpu-compat */
    if (qemuCaps->version >= 2010000 &&
        ARCH_IS_PPC64(qemuCaps->arch)) {
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_PSERIES_MAX_CPU_COMPAT);
    }

4267 4268
    if (virQEMUCapsProbeQMPCommands(qemuCaps, mon) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
4269 4270 4271 4272 4273

    /* Some capabilities may differ depending on KVM state */
    if (virQEMUCapsProbeQMPKVMState(qemuCaps, mon) < 0)
        goto cleanup;

4274 4275
    if (virQEMUCapsProbeQMPEvents(qemuCaps, mon) < 0)
        goto cleanup;
4276
    if (virQEMUCapsProbeQMPDevices(qemuCaps, mon) < 0)
4277 4278 4279
        goto cleanup;
    if (virQEMUCapsProbeQMPMachineTypes(qemuCaps, mon) < 0)
        goto cleanup;
4280
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, false) < 0)
4281 4282 4283 4284 4285
        goto cleanup;
    if (virQEMUCapsProbeQMPTPM(qemuCaps, mon) < 0)
        goto cleanup;
    if (virQEMUCapsProbeQMPCommandLine(qemuCaps, mon) < 0)
        goto cleanup;
4286 4287
    if (virQEMUCapsProbeQMPMigrationCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4288 4289 4290
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA) &&
        virQEMUCapsProbeQMPSchemaCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4291
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, false) < 0)
4292
        goto cleanup;
4293

4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304
    /* 'intel-iommu' shows up as a device since 2.2.0, but can
     * not be used with -device until 2.7.0. Before that it
     * requires -machine iommu=on. So we must clear the device
     * capability we detected on older QEMUs
     */
    if (qemuCaps->version < 2007000 &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_INTEL_IOMMU)) {
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_INTEL_IOMMU);
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_IOMMU);
    }

A
Andrea Bolognani 已提交
4305 4306
    /* GIC capabilities, eg. available GIC versions */
    if ((qemuCaps->arch == VIR_ARCH_AARCH64 ||
S
Stefan Schallenberg 已提交
4307
         qemuCaps->arch == VIR_ARCH_ARMV6L ||
A
Andrea Bolognani 已提交
4308 4309 4310 4311
         qemuCaps->arch == VIR_ARCH_ARMV7L) &&
        virQEMUCapsProbeQMPGICCapabilities(qemuCaps, mon) < 0)
        goto cleanup;

4312 4313 4314 4315 4316 4317 4318
    /* Prealloc on NVDIMMs is broken on older QEMUs leading to
     * user data corruption. If we are dealing with such version
     * of QEMU pretend we don't know how to NVDIMM. */
    if (qemuCaps->version < 2009000 &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_NVDIMM))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_NVDIMM);

4319 4320 4321 4322
    if (ARCH_IS_X86(qemuCaps->arch) &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_CPU_CACHE);

4323 4324 4325 4326 4327
    if (ARCH_IS_S390(qemuCaps->arch)) {
        /* Legacy assurance for QEMU_CAPS_CCW */
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW) &&
            virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_CCW))
            virQEMUCapsSet(qemuCaps, QEMU_CAPS_CCW);
4328 4329
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW_CSSID_UNRESTRICTED))
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_VFIO_CCW);
4330 4331
    }

4332 4333
    /* Probe for SEV capabilities */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST)) {
4334 4335 4336 4337 4338 4339
        int rc = virQEMUCapsProbeQMPSEVCapabilities(qemuCaps, mon);

        if (rc < 0)
            goto cleanup;

        if (rc == 0)
4340 4341 4342
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_SEV_GUEST);
    }

4343 4344 4345 4346 4347
    /* To avoid guest ABI regression, blockdev shall be enabled only when
     * we are able to pass the custom 'device_id' for SCSI disks and cdroms. */
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_DISK_DEVICE_ID))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_BLOCKDEV);

4348
    ret = 0;
4349
 cleanup:
4350 4351 4352
    return ret;
}

4353

4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366
int
virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps ATTRIBUTE_UNUSED,
                             qemuMonitorPtr mon)
{
    int ret = -1;

    if (qemuMonitorSetCapabilities(mon) < 0) {
        VIR_DEBUG("Failed to set monitor capabilities %s",
                  virGetLastErrorMessage());
        ret = 0;
        goto cleanup;
    }

4367 4368 4369
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, true) < 0)
        goto cleanup;

4370 4371 4372
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, true) < 0)
        goto cleanup;

4373 4374 4375 4376 4377 4378
    ret = 0;
 cleanup:
    return ret;
}


4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390
typedef struct _virQEMUCapsInitQMPCommand virQEMUCapsInitQMPCommand;
typedef virQEMUCapsInitQMPCommand *virQEMUCapsInitQMPCommandPtr;
struct _virQEMUCapsInitQMPCommand {
    char *binary;
    uid_t runUid;
    gid_t runGid;
    char **qmperr;
    char *monarg;
    char *monpath;
    char *pidfile;
    virCommandPtr cmd;
    qemuMonitorPtr mon;
4391
    virDomainChrSourceDef config;
4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409
    pid_t pid;
    virDomainObjPtr vm;
};


static void
virQEMUCapsInitQMPCommandAbort(virQEMUCapsInitQMPCommandPtr cmd)
{
    if (cmd->mon)
        virObjectUnlock(cmd->mon);
    qemuMonitorClose(cmd->mon);
    cmd->mon = NULL;

    virCommandAbort(cmd->cmd);
    virCommandFree(cmd->cmd);
    cmd->cmd = NULL;

    if (cmd->monpath)
4410
        unlink(cmd->monpath);
4411 4412 4413 4414 4415 4416

    virDomainObjEndAPI(&cmd->vm);

    if (cmd->pid != 0) {
        char ebuf[1024];

4417
        VIR_DEBUG("Killing QMP caps process %lld", (long long)cmd->pid);
4418 4419
        if (virProcessKill(cmd->pid, SIGKILL) < 0 && errno != ESRCH)
            VIR_ERROR(_("Failed to kill process %lld: %s"),
4420
                      (long long)cmd->pid,
4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463
                      virStrerror(errno, ebuf, sizeof(ebuf)));

        VIR_FREE(*cmd->qmperr);
    }
    if (cmd->pidfile)
        unlink(cmd->pidfile);
    cmd->pid = 0;
}


static void
virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd)
{
    if (!cmd)
        return;

    virQEMUCapsInitQMPCommandAbort(cmd);
    VIR_FREE(cmd->binary);
    VIR_FREE(cmd->monpath);
    VIR_FREE(cmd->monarg);
    VIR_FREE(cmd->pidfile);
    VIR_FREE(cmd);
}


static virQEMUCapsInitQMPCommandPtr
virQEMUCapsInitQMPCommandNew(char *binary,
                             const char *libDir,
                             uid_t runUid,
                             gid_t runGid,
                             char **qmperr)
{
    virQEMUCapsInitQMPCommandPtr cmd = NULL;

    if (VIR_ALLOC(cmd) < 0)
        goto error;

    if (VIR_STRDUP(cmd->binary, binary) < 0)
        goto error;

    cmd->runUid = runUid;
    cmd->runGid = runGid;
    cmd->qmperr = qmperr;
4464

4465 4466 4467
    /* the ".sock" sufix is important to avoid a possible clash with a qemu
     * domain called "capabilities"
     */
4468 4469 4470 4471 4472
    if (virAsprintf(&cmd->monpath, "%s/%s", libDir,
                    "capabilities.monitor.sock") < 0)
        goto error;
    if (virAsprintf(&cmd->monarg, "unix:%s,server,nowait", cmd->monpath) < 0)
        goto error;
4473

4474 4475
    /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash
     * with a qemu domain called "capabilities"
4476 4477 4478
     * Normally we'd use runDir for pid files, but because we're using
     * -daemonize we need QEMU to be allowed to create them, rather
     * than libvirtd. So we're using libDir which QEMU can write to
4479
     */
4480 4481
    if (virAsprintf(&cmd->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0)
        goto error;
4482

4483
    virPidFileForceCleanupPath(cmd->pidfile);
4484

4485 4486 4487
    cmd->config.type = VIR_DOMAIN_CHR_TYPE_UNIX;
    cmd->config.data.nix.path = cmd->monpath;
    cmd->config.data.nix.listen = false;
4488

4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501
    return cmd;

 error:
    virQEMUCapsInitQMPCommandFree(cmd);
    return NULL;
}


/* Returns -1 on fatal error,
 *          0 on success,
 *          1 when probing QEMU failed
 */
static int
4502 4503
virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd,
                             bool forceTCG)
4504 4505
{
    virDomainXMLOptionPtr xmlopt = NULL;
4506
    const char *machine;
4507 4508 4509
    int status = 0;
    int ret = -1;

4510 4511 4512 4513 4514 4515 4516
    if (forceTCG)
        machine = "none,accel=tcg";
    else
        machine = "none,accel=kvm:tcg";

    VIR_DEBUG("Try to probe capabilities of '%s' via QMP, machine %s",
              cmd->binary, machine);
4517

4518 4519 4520 4521 4522 4523 4524
    /*
     * We explicitly need to use -daemonize here, rather than
     * virCommandDaemonize, because we need to synchronize
     * with QEMU creating its monitor socket API. Using
     * daemonize guarantees control won't return to libvirt
     * until the socket is present.
     */
4525 4526 4527 4528 4529
    cmd->cmd = virCommandNewArgList(cmd->binary,
                                    "-S",
                                    "-no-user-config",
                                    "-nodefaults",
                                    "-nographic",
4530
                                    "-machine", machine,
4531 4532 4533 4534 4535 4536
                                    "-qmp", cmd->monarg,
                                    "-pidfile", cmd->pidfile,
                                    "-daemonize",
                                    NULL);
    virCommandAddEnvPassCommon(cmd->cmd);
    virCommandClearCaps(cmd->cmd);
4537 4538 4539 4540

#if WITH_CAPNG
    /* QEMU might run into permission issues, e.g. /dev/sev (0600), override
     * them just for the purpose of probing */
4541 4542
    if (geteuid() == 0)
        virCommandAllowCap(cmd->cmd, CAP_DAC_OVERRIDE);
4543 4544
#endif

4545 4546 4547 4548
    virCommandSetGID(cmd->cmd, cmd->runGid);
    virCommandSetUID(cmd->cmd, cmd->runUid);

    virCommandSetErrorBuffer(cmd->cmd, cmd->qmperr);
4549

4550
    /* Log, but otherwise ignore, non-zero status.  */
4551
    if (virCommandRun(cmd->cmd, &status) < 0)
4552 4553 4554
        goto cleanup;

    if (status != 0) {
4555
        VIR_DEBUG("QEMU %s exited with status %d: %s",
4556 4557
                  cmd->binary, status, *cmd->qmperr);
        goto ignore;
4558 4559
    }

4560 4561 4562
    if (virPidFileReadPath(cmd->pidfile, &cmd->pid) < 0) {
        VIR_DEBUG("Failed to read pidfile %s", cmd->pidfile);
        goto ignore;
4563 4564
    }

4565
    if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) ||
4566
        !(cmd->vm = virDomainObjNew(xmlopt)))
4567 4568
        goto cleanup;

4569
    cmd->vm->pid = cmd->pid;
4570

4571
    if (!(cmd->mon = qemuMonitorOpen(cmd->vm, &cmd->config, true, true,
4572
                                     0, &callbacks, NULL)))
4573
        goto ignore;
4574

4575
    virObjectLock(cmd->mon);
4576 4577 4578

    ret = 0;

4579
 cleanup:
4580 4581
    if (!cmd->mon)
        virQEMUCapsInitQMPCommandAbort(cmd);
4582
    virObjectUnref(xmlopt);
4583

4584
    return ret;
4585

4586 4587 4588 4589
 ignore:
    ret = 1;
    goto cleanup;
}
4590

4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606

static int
virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps,
                   const char *libDir,
                   uid_t runUid,
                   gid_t runGid,
                   char **qmperr)
{
    virQEMUCapsInitQMPCommandPtr cmd = NULL;
    int ret = -1;
    int rc;

    if (!(cmd = virQEMUCapsInitQMPCommandNew(qemuCaps->binary, libDir,
                                             runUid, runGid, qmperr)))
        goto cleanup;

4607
    if ((rc = virQEMUCapsInitQMPCommandRun(cmd, false)) != 0) {
4608 4609 4610
        if (rc == 1)
            ret = 0;
        goto cleanup;
4611
    }
4612 4613 4614 4615

    if (virQEMUCapsInitQMPMonitor(qemuCaps, cmd->mon) < 0)
        goto cleanup;

4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
        virQEMUCapsInitQMPCommandAbort(cmd);
        if ((rc = virQEMUCapsInitQMPCommandRun(cmd, true)) != 0) {
            if (rc == 1)
                ret = 0;
            goto cleanup;
        }

        if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, cmd->mon) < 0)
            goto cleanup;
    }

4628 4629 4630 4631
    ret = 0;

 cleanup:
    virQEMUCapsInitQMPCommandFree(cmd);
4632 4633 4634 4635
    return ret;
}


4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646
#define MESSAGE_ID_CAPS_PROBE_FAILURE "8ae2f3fb-2dbe-498e-8fbd-012d40afa361"

static void
virQEMUCapsLogProbeFailure(const char *binary)
{
    virLogMetadata meta[] = {
        { .key = "MESSAGE_ID", .s = MESSAGE_ID_CAPS_PROBE_FAILURE, .iv = 0 },
        { .key = "LIBVIRT_QEMU_BINARY", .s = binary, .iv = 0 },
        { .key = NULL },
    };

4647
    virLogMessage(&virLogSelf,
4648 4649 4650 4651
                  VIR_LOG_WARN,
                  __FILE__, __LINE__, __func__,
                  meta,
                  _("Failed to probe capabilities for %s: %s"),
4652
                  binary, virGetLastErrorMessage());
4653 4654 4655
}


4656
virQEMUCapsPtr
4657
virQEMUCapsNewForBinaryInternal(virArch hostArch,
4658
                                const char *binary,
4659 4660 4661
                                const char *libDir,
                                uid_t runUid,
                                gid_t runGid,
4662
                                unsigned int microcodeVersion,
J
Ján Tomko 已提交
4663
                                const char *kernelVersion)
4664
{
4665
    virQEMUCapsPtr qemuCaps;
4666
    struct stat sb;
4667
    char *qmperr = NULL;
4668

4669 4670 4671
    if (!(qemuCaps = virQEMUCapsNew()))
        goto error;

4672 4673
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;
4674 4675 4676 4677 4678 4679 4680 4681

    /* We would also want to check faccessat if we cared about ACLs,
     * but we don't.  */
    if (stat(binary, &sb) < 0) {
        virReportSystemError(errno, _("Cannot check QEMU binary %s"),
                             binary);
        goto error;
    }
4682
    qemuCaps->ctime = sb.st_ctime;
4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693

    /* Make sure the binary we are about to try exec'ing exists.
     * Technically we could catch the exec() failure, but that's
     * in a sub-process so it's hard to feed back a useful error.
     */
    if (!virFileIsExecutable(binary)) {
        virReportSystemError(errno, _("QEMU binary %s is not executable"),
                             binary);
        goto error;
    }

4694 4695
    if (virQEMUCapsInitQMP(qemuCaps, libDir, runUid, runGid, &qmperr) < 0) {
        virQEMUCapsLogProbeFailure(binary);
4696
        goto error;
4697
    }
4698

J
Ján Tomko 已提交
4699
    if (!qemuCaps->usedQMP) {
4700 4701 4702 4703 4704 4705
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to probe QEMU binary with QMP: %s"),
                       qmperr ? qmperr : _("unknown error"));
        virQEMUCapsLogProbeFailure(binary);
        goto error;
    }
4706

4707 4708
    qemuCaps->libvirtCtime = virGetSelfLastChanged();
    qemuCaps->libvirtVersion = LIBVIR_VERSION_NUMBER;
4709

4710 4711
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
4712

4713
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
4714 4715
        qemuCaps->microcodeVersion = microcodeVersion;

4716 4717
        if (VIR_STRDUP(qemuCaps->kernelVersion, kernelVersion) < 0)
            goto error;
4718 4719

        qemuCaps->kvmSupportsNesting = virQEMUCapsKVMSupportsNesting();
4720 4721
    }

4722
 cleanup:
4723
    VIR_FREE(qmperr);
4724
    return qemuCaps;
4725

4726
 error:
4727 4728
    virObjectUnref(qemuCaps);
    qemuCaps = NULL;
4729
    goto cleanup;
4730 4731
}

4732 4733 4734
static void *
virQEMUCapsNewData(const char *binary,
                   void *privData)
4735
{
4736 4737 4738 4739 4740 4741 4742
    virQEMUCapsCachePrivPtr priv = privData;

    return virQEMUCapsNewForBinaryInternal(priv->hostArch,
                                           binary,
                                           priv->libDir,
                                           priv->runUid,
                                           priv->runGid,
4743
                                           priv->microcodeVersion,
J
Ján Tomko 已提交
4744
                                           priv->kernelVersion);
4745
}
4746 4747


4748 4749 4750 4751 4752 4753 4754
static void *
virQEMUCapsLoadFile(const char *filename,
                    const char *binary,
                    void *privData)
{
    virQEMUCapsPtr qemuCaps = virQEMUCapsNew();
    virQEMUCapsCachePrivPtr priv = privData;
4755

4756 4757
    if (!qemuCaps)
        return NULL;
4758

4759 4760 4761 4762 4763 4764 4765
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;

    if (virQEMUCapsLoadCache(priv->hostArch, qemuCaps, filename) < 0)
        goto error;

 cleanup:
4766 4767 4768 4769
    return qemuCaps;

 error:
    virObjectUnref(qemuCaps);
4770 4771
    qemuCaps = NULL;
    goto cleanup;
4772 4773
}

4774

4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807
struct virQEMUCapsMachineTypeFilter {
    const char *machineType;
    virQEMUCapsFlags *flags;
    size_t nflags;
};

static const struct virQEMUCapsMachineTypeFilter virQEMUCapsMachineFilter[] = {
    /* { "blah", virQEMUCapsMachineBLAHFilter,
         ARRAY_CARDINALITY(virQEMUCapsMachineBLAHFilter) }, */
    { "", NULL, 0 },
};


void
virQEMUCapsFilterByMachineType(virQEMUCapsPtr qemuCaps,
                               const char *machineType)
{
    size_t i;

    if (!machineType)
        return;

    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsMachineFilter); i++) {
        const struct virQEMUCapsMachineTypeFilter *filter = &virQEMUCapsMachineFilter[i];
        size_t j;

        if (STRNEQ(filter->machineType, machineType))
            continue;

        for (j = 0; j < filter->nflags; j++)
            virQEMUCapsClear(qemuCaps, filter->flags[j]);
    }

4808 4809
    if (!virQEMUCapsGetMachineHotplugCpus(qemuCaps, machineType))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
4810 4811 4812
}


4813 4814 4815 4816 4817 4818 4819 4820 4821 4822
virFileCacheHandlers qemuCapsCacheHandlers = {
    .isValid = virQEMUCapsIsValid,
    .newData = virQEMUCapsNewData,
    .loadFile = virQEMUCapsLoadFile,
    .saveFile = virQEMUCapsSaveFile,
    .privFree = virQEMUCapsCachePrivFree,
};


virFileCachePtr
4823
virQEMUCapsCacheNew(const char *libDir,
4824
                    const char *cacheDir,
4825
                    uid_t runUid,
4826 4827
                    gid_t runGid,
                    unsigned int microcodeVersion)
4828
{
4829 4830 4831
    char *capsCacheDir = NULL;
    virFileCachePtr cache = NULL;
    virQEMUCapsCachePrivPtr priv = NULL;
4832
    struct utsname uts;
4833

4834
    if (virAsprintf(&capsCacheDir, "%s/capabilities", cacheDir) < 0)
4835
        goto error;
4836 4837

    if (!(cache = virFileCacheNew(capsCacheDir, "xml", &qemuCapsCacheHandlers)))
4838
        goto error;
4839

4840
    if (VIR_ALLOC(priv) < 0)
4841
        goto error;
4842
    virFileCacheSetPriv(cache, priv);
4843

4844
    if (VIR_STRDUP(priv->libDir, libDir) < 0)
4845 4846
        goto error;

4847
    priv->hostArch = virArchFromHost();
4848

4849 4850
    priv->runUid = runUid;
    priv->runGid = runGid;
4851
    priv->microcodeVersion = microcodeVersion;
4852
    priv->kvmUsable = VIR_TRISTATE_BOOL_ABSENT;
4853

4854 4855 4856 4857
    if (uname(&uts) == 0 &&
        virAsprintf(&priv->kernelVersion, "%s %s", uts.release, uts.version) < 0)
        goto error;

4858 4859
 cleanup:
    VIR_FREE(capsCacheDir);
4860 4861
    return cache;

4862
 error:
4863 4864 4865
    virObjectUnref(cache);
    cache = NULL;
    goto cleanup;
4866 4867 4868
}


4869
virQEMUCapsPtr
4870
virQEMUCapsCacheLookup(virFileCachePtr cache,
4871
                       const char *binary)
4872
{
4873
    virQEMUCapsPtr ret = NULL;
4874

4875
    ret = virFileCacheLookup(cache, binary);
4876 4877

    VIR_DEBUG("Returning caps %p for %s", ret, binary);
4878 4879 4880 4881
    return ret;
}


4882
virQEMUCapsPtr
4883
virQEMUCapsCacheLookupCopy(virFileCachePtr cache,
4884
                           const char *binary,
4885
                           const char *machineType)
4886
{
4887
    virQEMUCapsPtr qemuCaps = virQEMUCapsCacheLookup(cache, binary);
4888
    virQEMUCapsPtr ret;
4889

4890
    if (!qemuCaps)
4891 4892
        return NULL;

4893 4894
    ret = virQEMUCapsNewCopy(qemuCaps);
    virObjectUnref(qemuCaps);
4895 4896 4897 4898

    if (!ret)
        return NULL;

4899
    virQEMUCapsFilterByMachineType(ret, machineType);
4900 4901 4902 4903
    return ret;
}


4904 4905 4906 4907 4908
static int
virQEMUCapsCompareArch(const void *payload,
                       const void *name ATTRIBUTE_UNUSED,
                       const void *opaque)
{
4909
    struct virQEMUCapsSearchData *data = (struct virQEMUCapsSearchData *)opaque;
4910 4911
    const virQEMUCaps *qemuCaps = payload;

4912 4913 4914 4915 4916 4917 4918 4919 4920
    if (qemuCaps->arch != data->arch)
        return false;

    if (data->binaryFilter &&
        !strstr(qemuCaps->binary, data->binaryFilter)) {
        return false;
    }

    return true;
4921 4922 4923 4924
}


virQEMUCapsPtr
4925
virQEMUCapsCacheLookupByArch(virFileCachePtr cache,
4926 4927 4928
                             virArch arch)
{
    virQEMUCapsPtr ret = NULL;
4929 4930 4931 4932
    const char *binaryFilters[] = {
        "qemu-system-",
        NULL,
    };
4933 4934 4935 4936
    virArch archs[] = {
        arch,
        virQEMUCapsFindTarget(virArchFromHost(), arch),
    };
4937
    size_t i;
4938 4939
    size_t j;

4940 4941 4942 4943 4944 4945
    for (i = 0; i < ARRAY_CARDINALITY(binaryFilters); i++) {
        for (j = 0; j < ARRAY_CARDINALITY(archs); j++) {
            struct virQEMUCapsSearchData data = {
                .arch = archs[j],
                .binaryFilter = binaryFilters[i],
            };
4946

4947 4948 4949 4950
            ret = virFileCacheLookupByFunc(cache, virQEMUCapsCompareArch, &data);
            if (ret)
                goto done;
        }
4951 4952
    }

4953 4954 4955 4956 4957
    virReportError(VIR_ERR_INVALID_ARG,
                   _("unable to find any emulator to serve '%s' "
                     "architecture"), virArchToString(arch));

 done:
4958 4959
    VIR_DEBUG("Returning caps %p for arch %s", ret, virArchToString(arch));

4960 4961 4962 4963
    return ret;
}


4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049
/**
 * virQEMUCapsCacheLookupDefault:
 * @cache: QEMU capabilities cache
 * @binary: optional path to QEMU binary
 * @archStr: optional guest architecture
 * @virttypeStr: optional virt type
 * @machine: optional machine type
 * @retArch: if non-NULL, guest architecture will be returned here
 * @retVirttype: if non-NULL, domain virt type will be returned here
 * @retMachine: if non-NULL, canonical machine type will be returned here
 *
 * Looks up the QEMU binary specified by @binary and @archStr, checks it can
 * provide the required @virttypeStr and @machine and returns its capabilities.
 * Sensible defaults are used for any argument which is NULL (the function can
 * even be called with all NULL arguments).
 *
 * Returns QEMU capabilities matching the requirements, NULL on error.
 */
virQEMUCapsPtr
virQEMUCapsCacheLookupDefault(virFileCachePtr cache,
                              const char *binary,
                              const char *archStr,
                              const char *virttypeStr,
                              const char *machine,
                              virArch *retArch,
                              virDomainVirtType *retVirttype,
                              const char **retMachine)
{
    int virttype = VIR_DOMAIN_VIRT_NONE;
    int arch = virArchFromHost();
    virDomainVirtType capsType;
    virQEMUCapsPtr qemuCaps = NULL;
    virQEMUCapsPtr ret = NULL;

    if (virttypeStr &&
        (virttype = virDomainVirtTypeFromString(virttypeStr)) < 0) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("unknown virttype: %s"), virttypeStr);
        goto cleanup;
    }

    if (archStr &&
        (arch = virArchFromString(archStr)) == VIR_ARCH_NONE) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("unknown architecture: %s"), archStr);
        goto cleanup;
    }

    if (binary) {
        virArch arch_from_caps;

        if (!(qemuCaps = virQEMUCapsCacheLookup(cache, binary)))
            goto cleanup;

        arch_from_caps = virQEMUCapsGetArch(qemuCaps);

        if (arch_from_caps != arch &&
            !((ARCH_IS_X86(arch) && ARCH_IS_X86(arch_from_caps)) ||
              (ARCH_IS_PPC(arch) && ARCH_IS_PPC(arch_from_caps)) ||
              (ARCH_IS_ARM(arch) && ARCH_IS_ARM(arch_from_caps)) ||
              (ARCH_IS_S390(arch) && ARCH_IS_S390(arch_from_caps)))) {
            virReportError(VIR_ERR_INVALID_ARG,
                           _("architecture from emulator '%s' doesn't "
                             "match given architecture '%s'"),
                           virArchToString(arch_from_caps),
                           virArchToString(arch));
            goto cleanup;
        }
    } else {
        if (!(qemuCaps = virQEMUCapsCacheLookupByArch(cache, arch)))
            goto cleanup;

        binary = virQEMUCapsGetBinary(qemuCaps);
    }

    if (machine) {
        /* Turn @machine into canonical name */
        machine = virQEMUCapsGetCanonicalMachine(qemuCaps, machine);

        if (!virQEMUCapsIsMachineSupported(qemuCaps, machine)) {
            virReportError(VIR_ERR_INVALID_ARG,
                           _("the machine '%s' is not supported by emulator '%s'"),
                           machine, binary);
            goto cleanup;
        }
    } else {
5050
        machine = virQEMUCapsGetPreferredMachine(qemuCaps);
5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081
    }

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM))
        capsType = VIR_DOMAIN_VIRT_KVM;
    else
        capsType = VIR_DOMAIN_VIRT_QEMU;

    if (virttype == VIR_DOMAIN_VIRT_NONE)
        virttype = capsType;

    if (virttype == VIR_DOMAIN_VIRT_KVM && capsType == VIR_DOMAIN_VIRT_QEMU) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("KVM is not supported by '%s' on this host"),
                       binary);
        goto cleanup;
    }

    if (retArch)
        *retArch = arch;
    if (retVirttype)
        *retVirttype = virttype;
    if (retMachine)
        *retMachine = machine;

    VIR_STEAL_PTR(ret, qemuCaps);

 cleanup:
    virObjectUnref(qemuCaps);
    return ret;
}

5082 5083 5084 5085 5086 5087 5088
bool
virQEMUCapsSupportsVmport(virQEMUCapsPtr qemuCaps,
                          const virDomainDef *def)
{
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT))
        return false;

5089 5090
    return qemuDomainIsI440FX(def) ||
        qemuDomainIsQ35(def) ||
5091 5092 5093 5094
        STREQ(def->os.machine, "isapc");
}


5095 5096 5097 5098 5099 5100 5101
bool
virQEMUCapsIsMachineSupported(virQEMUCapsPtr qemuCaps,
                              const char *canonical_machine)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
5102
        if (STREQ(canonical_machine, qemuCaps->machineTypes[i].name))
5103 5104 5105 5106
            return true;
    }
    return false;
}
5107 5108


5109 5110 5111 5112
/*
 * The preferred machine to use if none is listed explicitly
 * Note that this may differ from QEMU's own default machine
 */
5113
const char *
5114
virQEMUCapsGetPreferredMachine(virQEMUCapsPtr qemuCaps)
5115 5116 5117
{
    if (!qemuCaps->nmachineTypes)
        return NULL;
5118
    return qemuCaps->machineTypes[0].name;
5119
}
5120 5121


5122
static int
5123
virQEMUCapsFillDomainLoaderCaps(virDomainCapsLoaderPtr capsLoader,
5124 5125
                                virFirmwarePtr *firmwares,
                                size_t nfirmwares)
5126
{
5127 5128
    size_t i;

5129
    capsLoader->supported = true;
5130

5131
    if (VIR_ALLOC_N(capsLoader->values.values, nfirmwares) < 0)
5132 5133
        return -1;

5134 5135
    for (i = 0; i < nfirmwares; i++) {
        const char *filename = firmwares[i]->name;
5136 5137 5138 5139 5140 5141

        if (!virFileExists(filename)) {
            VIR_DEBUG("loader filename=%s does not exist", filename);
            continue;
        }

5142
        if (VIR_STRDUP(capsLoader->values.values[capsLoader->values.nvalues],
5143 5144
                       filename) < 0)
            return -1;
5145
        capsLoader->values.nvalues++;
5146 5147
    }

5148
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
5149 5150
                             VIR_DOMAIN_LOADER_TYPE_ROM);

5151 5152
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
                             VIR_DOMAIN_LOADER_TYPE_PFLASH);
5153 5154


5155 5156 5157
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->readonly,
                             VIR_TRISTATE_BOOL_YES,
                             VIR_TRISTATE_BOOL_NO);
5158
    return 0;
5159 5160 5161
}


5162
static int
5163
virQEMUCapsFillDomainOSCaps(virDomainCapsOSPtr os,
5164 5165
                            virFirmwarePtr *firmwares,
                            size_t nfirmwares)
5166
{
5167
    virDomainCapsLoaderPtr capsLoader = &os->loader;
5168

5169
    os->supported = true;
5170
    if (virQEMUCapsFillDomainLoaderCaps(capsLoader, firmwares, nfirmwares) < 0)
5171 5172
        return -1;
    return 0;
5173 5174 5175
}


5176 5177 5178 5179 5180
static int
virQEMUCapsFillDomainCPUCaps(virCapsPtr caps,
                             virQEMUCapsPtr qemuCaps,
                             virDomainCapsPtr domCaps)
{
5181 5182
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_HOST_PASSTHROUGH))
5183 5184
        domCaps->cpu.hostPassthrough = true;

5185
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
5186
                                      VIR_CPU_MODE_HOST_MODEL)) {
5187 5188
        virCPUDefPtr cpu = virQEMUCapsGetHostModel(qemuCaps, domCaps->virttype,
                                                   VIR_QEMU_CAPS_HOST_CPU_REPORTED);
5189 5190
        domCaps->cpu.hostModel = virCPUDefCopy(cpu);
    }
5191 5192 5193 5194 5195

    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_CUSTOM)) {
        virDomainCapsCPUModelsPtr filtered = NULL;
        char **models = NULL;
5196
        const char *blacklist[] = { "host", NULL };
5197

J
Jiri Denemark 已提交
5198
        if (virCPUGetModels(domCaps->arch, &models) >= 0) {
5199 5200 5201 5202 5203 5204 5205 5206
            virDomainCapsCPUModelsPtr cpus;

            if (domCaps->virttype == VIR_DOMAIN_VIRT_KVM)
                cpus = qemuCaps->kvmCPUModels;
            else
                cpus = qemuCaps->tcgCPUModels;

            filtered = virDomainCapsCPUModelsFilter(cpus,
5207
                                                    (const char **)models,
5208
                                                    blacklist);
5209
            virStringListFree(models);
5210 5211
        }
        domCaps->cpu.custom = filtered;
5212
    }
5213 5214 5215 5216 5217

    return 0;
}


5218 5219 5220 5221 5222 5223 5224 5225 5226 5227
static int
virQEMUCapsFillDomainIOThreadCaps(virQEMUCapsPtr qemuCaps,
                                  virDomainCapsPtr domCaps)
{
    domCaps->iothreads = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD);

    return 0;
}


5228
static int
5229
virQEMUCapsFillDomainDeviceDiskCaps(virQEMUCapsPtr qemuCaps,
5230
                                    const char *machine,
5231 5232
                                    virDomainCapsDeviceDiskPtr disk)
{
5233
    disk->supported = true;
5234 5235 5236
    /* QEMU supports all of these */
    VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice,
                             VIR_DOMAIN_DISK_DEVICE_DISK,
5237 5238
                             VIR_DOMAIN_DISK_DEVICE_CDROM,
                             VIR_DOMAIN_DISK_DEVICE_LUN);
5239 5240

    /* PowerPC pseries based VMs do not support floppy device */
5241
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch)) {
5242
        VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice, VIR_DOMAIN_DISK_DEVICE_FLOPPY);
5243 5244
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_FDC);
    }
5245

5246 5247 5248
    if (qemuDomainMachineHasBuiltinIDE(machine))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_IDE);

5249 5250 5251 5252 5253 5254 5255
    VIR_DOMAIN_CAPS_ENUM_SET(disk->bus,
                             VIR_DOMAIN_DISK_BUS_SCSI,
                             VIR_DOMAIN_DISK_BUS_VIRTIO,
                             /* VIR_DOMAIN_DISK_BUS_SD */);

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_USB_STORAGE))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_USB);
5256 5257 5258 5259

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_AHCI))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_SATA);

5260
    return 0;
5261 5262 5263
}


5264 5265 5266 5267 5268 5269
static int
virQEMUCapsFillDomainDeviceGraphicsCaps(virQEMUCapsPtr qemuCaps,
                                        virDomainCapsDeviceGraphicsPtr dev)
{
    dev->supported = true;

J
Ján Tomko 已提交
5270
    VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_SDL);
5271 5272 5273 5274 5275 5276 5277 5278 5279
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VNC))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_VNC);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPICE))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_SPICE);

    return 0;
}


5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291
static int
virQEMUCapsFillDomainDeviceVideoCaps(virQEMUCapsPtr qemuCaps,
                                     virDomainCapsDeviceVideoPtr dev)
{
    dev->supported = true;

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VGA))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_VGA);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_CIRRUS_VGA))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_CIRRUS);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMWARE_SVGA))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_VMVGA);
5292
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QXL))
5293 5294 5295 5296 5297 5298 5299 5300
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_QXL);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_GPU))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_VIRTIO);

    return 0;
}


5301
static int
5302 5303 5304 5305 5306 5307
virQEMUCapsFillDomainDeviceHostdevCaps(virQEMUCapsPtr qemuCaps,
                                       virDomainCapsDeviceHostdevPtr hostdev)
{
    bool supportsPassthroughKVM = qemuHostdevHostSupportsPassthroughLegacy();
    bool supportsPassthroughVFIO = qemuHostdevHostSupportsPassthroughVFIO();

5308
    hostdev->supported = true;
5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320
    /* VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES is for containers only */
    VIR_DOMAIN_CAPS_ENUM_SET(hostdev->mode,
                             VIR_DOMAIN_HOSTDEV_MODE_SUBSYS);

    VIR_DOMAIN_CAPS_ENUM_SET(hostdev->startupPolicy,
                             VIR_DOMAIN_STARTUP_POLICY_DEFAULT,
                             VIR_DOMAIN_STARTUP_POLICY_MANDATORY,
                             VIR_DOMAIN_STARTUP_POLICY_REQUISITE,
                             VIR_DOMAIN_STARTUP_POLICY_OPTIONAL);

    VIR_DOMAIN_CAPS_ENUM_SET(hostdev->subsysType,
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB,
5321 5322
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI,
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI);
5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334

    /* No virDomainHostdevCapsType for QEMU */
    virDomainCapsEnumClear(&hostdev->capsType);

    virDomainCapsEnumClear(&hostdev->pciBackend);
    if (supportsPassthroughVFIO &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VFIO_PCI)) {
        VIR_DOMAIN_CAPS_ENUM_SET(hostdev->pciBackend,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO);
    }

J
Ján Tomko 已提交
5335
    if (supportsPassthroughKVM) {
5336 5337 5338 5339
        VIR_DOMAIN_CAPS_ENUM_SET(hostdev->pciBackend,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM);
    }
5340
    return 0;
5341 5342 5343
}


5344 5345 5346 5347 5348 5349 5350
/**
 * virQEMUCapsSupportsGICVersion:
 * @qemuCaps: QEMU capabilities
 * @virtType: domain type
 * @version: GIC version
 *
 * Checks the QEMU binary with capabilities @qemuCaps supports a specific
5351 5352
 * GIC version for a domain of type @virtType. If @qemuCaps is NULL, the GIC
 * @version is considered unsupported.
5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363
 *
 * Returns: true if the binary supports the requested GIC version, false
 *          otherwise
 */
bool
virQEMUCapsSupportsGICVersion(virQEMUCapsPtr qemuCaps,
                              virDomainVirtType virtType,
                              virGICVersion version)
{
    size_t i;

5364 5365 5366
    if (!qemuCaps)
        return false;

5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385
    for (i = 0; i < qemuCaps->ngicCapabilities; i++) {
        virGICCapabilityPtr cap = &(qemuCaps->gicCapabilities[i]);

        if (cap->version != version)
            continue;

        if (virtType == VIR_DOMAIN_VIRT_KVM &&
            cap->implementation & VIR_GIC_IMPLEMENTATION_KERNEL)
            return true;

        if (virtType == VIR_DOMAIN_VIRT_QEMU &&
            cap->implementation & VIR_GIC_IMPLEMENTATION_EMULATED)
            return true;
    }

    return false;
}


5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409
/**
 * virQEMUCapsFillDomainFeatureGICCaps:
 * @qemuCaps: QEMU capabilities
 * @domCaps: domain capabilities
 *
 * Take the information about GIC capabilities that has been obtained
 * using the 'query-gic-capabilities' QMP command and stored in @qemuCaps
 * and convert it to a form suitable for @domCaps.
 *
 * @qemuCaps contains complete information about the GIC capabilities for
 * the corresponding QEMU binary, stored as custom objects; @domCaps, on
 * the other hand, should only contain information about the GIC versions
 * available for the specific combination of architecture, machine type
 * and virtualization type. Moreover, a common format is used to store
 * information about enumerations in @domCaps, so further processing is
 * required.
 *
 * Returns: 0 on success, <0 on failure
 */
static int
virQEMUCapsFillDomainFeatureGICCaps(virQEMUCapsPtr qemuCaps,
                                    virDomainCapsPtr domCaps)
{
    virDomainCapsFeatureGICPtr gic = &domCaps->gic;
5410
    virGICVersion version;
5411

5412
    if (!qemuDomainMachineIsARMVirt(domCaps->machine, domCaps->arch))
5413 5414
        return 0;

5415 5416 5417 5418 5419 5420
    for (version = VIR_GIC_VERSION_LAST - 1;
         version > VIR_GIC_VERSION_NONE;
         version--) {
        if (!virQEMUCapsSupportsGICVersion(qemuCaps,
                                           domCaps->virttype,
                                           version))
5421 5422 5423 5424
            continue;

        gic->supported = true;
        VIR_DOMAIN_CAPS_ENUM_SET(gic->version,
5425
                                 version);
5426 5427 5428 5429 5430 5431
    }

    return 0;
}


5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447
/**
 * virQEMUCapsFillDomainFeatureSEVCaps:
 * @qemuCaps: QEMU capabilities
 * @domCaps: domain capabilities
 *
 * Take the information about SEV capabilities that has been obtained
 * using the 'query-sev-capabilities' QMP command and stored in @qemuCaps
 * and convert it to a form suitable for @domCaps.
 *
 * Returns: 0 on success, -1 on failure
 */
static int
virQEMUCapsFillDomainFeatureSEVCaps(virQEMUCapsPtr qemuCaps,
                                    virDomainCapsPtr domCaps)
{
    virSEVCapability *cap = qemuCaps->sevCapabilities;
5448
    VIR_AUTOPTR(virSEVCapability) sev = NULL;
5449 5450 5451 5452 5453 5454 5455 5456

    if (!cap)
        return 0;

    if (VIR_ALLOC(sev) < 0)
        return -1;

    if (VIR_STRDUP(sev->pdh, cap->pdh) < 0)
5457
        return -1;
5458 5459

    if (VIR_STRDUP(sev->cert_chain, cap->cert_chain) < 0)
5460
        return -1;
5461 5462 5463 5464 5465

    sev->cbitpos = cap->cbitpos;
    sev->reduced_phys_bits = cap->reduced_phys_bits;
    VIR_STEAL_PTR(domCaps->sev, sev);

5466
    return 0;
5467 5468 5469
}


5470
int
5471 5472
virQEMUCapsFillDomainCaps(virCapsPtr caps,
                          virDomainCapsPtr domCaps,
5473
                          virQEMUCapsPtr qemuCaps,
5474
                          virFirmwarePtr *firmwares,
5475
                          size_t nfirmwares)
5476
{
5477
    virDomainCapsOSPtr os = &domCaps->os;
5478 5479
    virDomainCapsDeviceDiskPtr disk = &domCaps->disk;
    virDomainCapsDeviceHostdevPtr hostdev = &domCaps->hostdev;
5480
    virDomainCapsDeviceGraphicsPtr graphics = &domCaps->graphics;
5481
    virDomainCapsDeviceVideoPtr video = &domCaps->video;
5482

5483 5484
    domCaps->maxvcpus = virQEMUCapsGetMachineMaxCpus(qemuCaps,
                                                     domCaps->machine);
5485
    if (domCaps->virttype == VIR_DOMAIN_VIRT_KVM) {
5486 5487 5488 5489 5490 5491
        int hostmaxvcpus;

        if ((hostmaxvcpus = virHostCPUGetKVMMaxVCPUs()) < 0)
            return -1;

        domCaps->maxvcpus = MIN(domCaps->maxvcpus, hostmaxvcpus);
5492
    }
5493

5494 5495 5496
    domCaps->vmcoreinfo = virQEMUCapsGet(qemuCaps,
                                         QEMU_CAPS_DEVICE_VMCOREINFO);

5497 5498
    domCaps->genid = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMGENID);

5499
    if (virQEMUCapsFillDomainOSCaps(os, firmwares, nfirmwares) < 0 ||
5500
        virQEMUCapsFillDomainCPUCaps(caps, qemuCaps, domCaps) < 0 ||
5501
        virQEMUCapsFillDomainIOThreadCaps(qemuCaps, domCaps) < 0 ||
5502 5503 5504
        virQEMUCapsFillDomainDeviceDiskCaps(qemuCaps,
                                            domCaps->machine, disk) < 0 ||
        virQEMUCapsFillDomainDeviceGraphicsCaps(qemuCaps, graphics) < 0 ||
5505
        virQEMUCapsFillDomainDeviceVideoCaps(qemuCaps, video) < 0 ||
5506
        virQEMUCapsFillDomainDeviceHostdevCaps(qemuCaps, hostdev) < 0 ||
5507 5508
        virQEMUCapsFillDomainFeatureGICCaps(qemuCaps, domCaps) < 0 ||
        virQEMUCapsFillDomainFeatureSEVCaps(qemuCaps, domCaps) < 0)
5509
        return -1;
5510

5511
    return 0;
5512
}
5513 5514 5515 5516 5517 5518 5519 5520


void
virQEMUCapsSetMicrocodeVersion(virQEMUCapsPtr qemuCaps,
                               unsigned int microcodeVersion)
{
    qemuCaps->microcodeVersion = microcodeVersion;
}
5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537


/**
 * virQEMUCapsStripMachineAliases:
 * @qemuCaps: capabilities object to process
 *
 * Remove all aliases so that the tests depending on the latest capabilities
 * file can be stable when new files are added.
 */
void
virQEMUCapsStripMachineAliases(virQEMUCapsPtr qemuCaps)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++)
        VIR_FREE(qemuCaps->machineTypes[i].alias);
}